FORMATION AND MODIFIED BLOOD FLOW OF ABDOMINAL AORTIC ANEURYSMS

Brandon Dizon, Kari Harrison, Wen Tsai MAE C150G Fluid Dynamics of Biological Systems Professor Eldredge Spring 2010

TABLE OF CONTENTS ABSTRACT INTRODUCTION LITERATURE REVIEW PROBLEM STATEMENT METHODOLOGY RESULT CONCLUSION BIBLIOGRAPHY 2 3 4 12 13 15 22 23

ABSTRACT In this study, four 2D models of a blood vessel with an aneurysm were created using COMSOLs modeling tool. The purpose of our study is to obtain insight into the blood flow behavior in an abdominal aneurysm and what may cause the aneurysm to rupture. When the size of aneurysm increases, the wall thickness decreases while the wall tension increases. A symmetrical two-dimensional model represents a straight vessel and a semicircular bulge represents the aneurysm. In the model, one can easily see that the flow is slower in the aneurysm area and the maximum speed occurs at the inlet of the vessel and at the ends of the aneurysm. Vortices are formed at the proximal and distal ends of the aneurysm. From the plots, one can see that as the radius of the aneurysm increases, the peak pressure on the wall increase. Overall, as the aneurysm gets larger, the wall thickness and the pressure on the wall should be considered.

INTRODUCTION Aneurysm is an area of a localized widening of a blood vessel. An aneurysm that forms in the belly area is called an abdominal aortic aneurysm (AAA) and it is the most common type of aortic aneurysm. There are four possible mechanisms of the formation: proteolytic degradation of aortic wall connective tissue, inflammation and immune responses, biochemical wall stress, and molecular genetics; however, the real reason is still unclear. An abdominal aortic aneurysm develops slowly over years and often has no symptoms. Until it ruptures or blood leaks out, one may feel pain in the abdomen or back, clammy skin, nausea and vomiting, rapid heart rate, or shock. It can lead to death within minutes if a large amount of blood spills into the abdominal area. According to emedicinehealth.com, ruptured abdominal aneurysms are the 13th leading cause of death in United States. Usually, abdominal aortic aneurysms happen to people between 65 to 70 years old, but people who have high blood pressure, high cholesterol, emphysema, genetric factors, smoke, or obesity may have a higher risk. In general, males have a higher possibility of getting AAA than females. This report compares the results of two-dimensional flow through a vessel with a bulge to simulate an aneuryzed blood vessel using CFD simulation with COMSOL. Additionally, this report covers several literature views.

Figure 1: The body

LITERATURE REVIEW

Of the several articles on abdominal aortic aneurysms, most focus on one parameter, such as symmetry, wall thickness, turbulence within the aneurysm, etc. Almost all are based purely on theoretical computational analysis using computer software, rather than scientific observation and experimentation, since this is incredibly difficult given todays technology and what is known about aneurysms. Scotti CM, Jimenez J, Muluck SC and Finol EA wrote two papers on the effects of asymmetrical aneurysm growth as well as wall thickness of the aorta. With this information, and further study, it may be possible to determine the causes of aneurysm rupture. Using CAD software, 10 virtual aneurysm models were created. All had varying asymmetry. 5 of the models had uniform wall thickness. The other 5 had variable wall thickness. With the capabilities of the CAD software they were able to analyze and predict the Von Mises stress according to the Huber-Von Mises-Hency theory (Scotti, et al 2005). It was concluded that as the aneurysm grew asymmetrically, the velocity vectors changed dramatically. A symmetric aneurysm had vortices that developed and dissipated within the midsection to distal end of the aneurysm sac. As asymmetry increased, the vortices traveled upstream to the proximal end. This decelerates the flow through the aneurysm and allows the vortices to remain longer along the anterior wall. Vortices within an aneurysm increase the amount of platelet deposition and increase the stress and risk of
Figure 2: Velocity vectors and fluid pressure for symmetric (top) and asymmetric (bottom) aneurysms

rupture. The simulations showed that the symmetric aneurysm had 14% lower Von Mises stresses than the most asymmetric aneurysm. Also, the location of the stress changed as asymmetry increased. In most aneurysm models, a uniform wall thickness is assumed to simplify the problems and calculations. However, experiments have shown that the wall actually thins as the aneurysm

expands. When compared to the variable wall stress model, uniform wall thickness underestimated the maximum stress by as much as 77%. There is also a difference in the location of this maximum stress. When the wall thickness is uniform, it occurs near the proximal and distal ends of the aneurysm. For the variable thickness model the maximum stress occurs at the midsection of the sac, where the wall is thinnest. It was concluded that the wall thickness variation presented in the models had a larger effect on wall stress than the asymmetry of the aneurysm (Scotti, et al 2005). There are some limitations of this model. First and foremost, wall stress is not the only cause of rupture, yet this is the only one explored. Secondly, it was assumed that the wall thickness decreased uniformly and continually toward the center of the aneurysm sac and he increased at the same rate towards the iliac bifurcation. It is unlikely that this is actually the case (Scotti, et al 2005).

Figure 3: Effect of uniform and variable wall thickness on velocity vectors (top) and stress distributions (bottom)

In an article by Hoi Y, Meng H, Woodward Sh, Bendolk BR, Hanel RA, Guterman RL, and Hopkins LN, the effects of arterial geometry were examined. Using a computational fluid dynamic analysis, the hemodynamics of an aneurysm from the lateral wall of the arties was studied. The arterial curves and neck sizes were varied and the effect of this on hemodynamic parameters was quantified. This was accomplished by using blood flow inertia as the predominate

force for driving blood into the aneurysms on curved arteries instead of the traditional, viscous diffusion method (Hoi, et al 2004). For simplification, incompressible and laminar flows were assumed and non Newtonian effects were neglected because the model consists of a large vessel with low shear rates. Viscosity was assumed to by 3.5 cP and density 1060 kg/m3. A no slip condition was applied to the walls and the wall was assumed to be rigid. Steady state pulsatile flow was simulated in 10 different models (Hoi, et al 2004). The area of the aneurysm wall where the wall shear stress was elevated to greater than 20 dynes/cm2 was calculated. This is the most likely site of active remodeling of the aneurysm wall based on previous studies. This is what is considered the
Figure 4: Graphs of impact zone size versus

impact zone. This zone was quantified in puslatile flow

simulations using time dependant wall shear stress distributions (Hoi, et al 2004). It was determined that all blood particle paths that entered the aneurysm did so at the distal side of the neck. Some paths didnt enter the aneurysm at all. Of those that entered, some left at the proximal side of the neck with a decelerated velocity, while others reentered the cavity and whirled irregularly. As curvature increases, so does flow impingement intensity. Also with increasing curvature the centrifugal forces on the outer wall of the aneurysm increased. The impact zone increased linearly with artery curvature and the third order of neck size (Hoi, et al 2004). The simulation results show that aneurysms on curved arteries were subjected to higher hemodynamic stresses. In addition, aneurysms with wider necks had larger impact zones. In correlation with findings from other studies; these impact zones are the most likely site for growth (Hoi, et al 2004).

Figure 6: Effect of increasing curvature and increasing neck size

Figure 7: Effect of increasing neck size and curvature on wall stress

The effect of turbulence inside a rigid aortic aneurysm was studied by Yip TH and Yu. Previous studies prove turbulence exists insides aneurysms of living patients. This model includes an investigation of the hydrodynamic stability of cyclic flow inside an aneurysm as well as a discussion of the mechanisms for the transition to turbulences for different stability characteristics. This is done by using a rectified sine wave to simulate aortic flow conditions, and obtaining stability characteristics from linear stability analysis on an unsteady velocity profile at different phases. Using living volunteers and an MRI machine, the velocity profile of a healthy aorta was modeled. Then, using a pipe network and oscillatory flow generated by a linear reciprocating piston driven by a motor, the experiment was developed (Yip, et al 2001).
Figure 7: Pipe network setup

In the pipe network experiment, a rigid glass aneurysm model was used. Two different models were used, one simulating an early stage aneurysm, the others simulating an advanced stage aneurysm. On top of this, different types of conditions were tested. In the first, flow remained laminar throughout the cycle. In the second, the flow was transition, with bursts of

turbulence. Velocity measurements were taken at between 228 and 276 locations. Using nylon particles, the flow patterns were photographed normal to the symmetry plane at different phases (Yip, et al 2001). It was found that as flow entered the aneurysm, vortices were formed at the proximal side and grew as they progressed along the wall of the sac. When t/T ~ 0.18, the vortices detached from the pipe wall and grew larger as they traveled toward the distal end. At t/T ~0.40, the strength of the vortices dissipated to near non existence at the distal end (Yip, et al 2001). When the phase angle = 33, a re-circulating vortex begins to form and expands as it rolls along the wall. It can detach from the wall once a certain momentum is achieved and is then in the bulk flow, causing turbulence at the distal neck. By = 171, vortecial structures completely vanish. Therefore, about 70% of the time, a vortex was pressed against the wall around the distal neck. Turbulence at the neck was only measured from = 93-135, and was at a maximum for 8% of its existence (Yip, et al 2001).

Figure 8: Vortex development for (a) t/T = 0.12, (b) t/T = 0.16, Figure 9: Phase variation of streamwise velocity for (c) t/T = 0.17, (d) t/T= 0.18, (e) t/T = 0.20, (f) t/T = 0.40 (a) 33 degrees, (b) 75 degrees, (d) 171 degrees

The previous models were all based on theoretical aorta and aneurysm properties. Wolters, Rutten, Schurink, Kose, de Hart and van de Vosses created a model for patient specific fluid structure interactions in abdominal aortic aneurysms. This was done in order to better predict the possibility of rupture. Using patient specific geometry acquired from CT angiographic images of aneurysms, a patient-specific hexahedral finite element mesh of the aneurysm was created and then simulations were used to determine the wall stress, blood velocity distribution and wall shear stress (Wolters, et al 2005).

The boundary surfaces and centerlines were determined using 3D Active Object. Then, a standardized bifurcation mesh is transformed based on the information and continuity and
Figure 10 Segmentation of aneurysm using (a) 3D Active Object: centerlines,

motion equations were used to compute the stress and strain

distributions. Based on the determined centerlines, the fluid domain of the mesh is transformed into an initial shape that is patient-specific and oriented along the lumen of the aneurysm. The fluid domain is then transformed to a patient-specific mesh based on local coordinate systems defined along the centerline. Finally, Laplacian smoothing is applied to correct for distorted elements (Wolters, et al 2005). The wall deformation problem is solved first, in order to determine the time varying displacement of the wall mesh. Next, the time varying wall stress distribution is determined. This is accomplished using wall constitutive behavior. Solving a linearly elastic deformation problem on the fluid domain gives the
Figure 11: Mesh generation procedure (a) standardized bifurcated mesh consisting of both fluid and solid domains (b) boundary surface,

motion of the fluid mesh. Then, the blood flow problem is solved in order to determine the blood velocity and pressure fields with respect to the deforming fluids mesh. The time varying wall shear stress distribution can be computed from the velocity field using the blood viscosity model (Wolters, et al 2005). It was determined that the wall stress distribution was characterized by the increase in stress around the bulging aneurysm, as well as local areas of increased and decreased stress distributions. At the apex of the bifurcation inner wall is where the highest stresses occurred. It was also determined that the stress to the anterior wall and posterior wall was related, although they differed in quantity. Consecutive formations of 3D vortices were observed. However, by the end-diastole there were few remnants of the vortices from the previous cycle. Finally, locally occurring vortices, as well as the local diameter determined a pattern of high and low wall shear stress (Wolters, et al 2005).

PROBLEM STATEMENT The aim of our simulation was to understand overall flow behavior through an aneurysm in the abdominal aorta. We will vary the size of the aneurismal sac and observe changes in flow patterns such as velocity flow vectors and formation of vortices. We will also calculate the force on the aneurysm wall due to viscous effects and pressure. The main goals of our model are the following: 1) Observe the fluid dynamics of blood flow through an abdominal aortic aneurysm and compare results to other CFD model results. 2) Use the CFD software to calculate pressure along the vessel walls. 3) Determine what role the size of aneurysm plays in blood flow through the vessel.

METHODOLOGY From our research we found that as an aneurysm increases in size, the walls of the blood vessel decrease in thickness (Scott et al.). The coupling of aneurismal dilation and increased wall tension can be approximated by Laplaces Law applied to cylinder (Arroyo & Lee, 1998). The law states that the circumferential stress, where is the pressure within the vessel, , can be approximated by the equation is the radius of the vessel and is the thickness of .

the vessel walls. In our model we will vary the value of

and use the software to calculate

In our simulation, we assumed that blood behaved as a Newtonian fluid. This assumption is reasonable in our model since the aorta is much larger in diameter (~3-4cm) than that of a red blood cell (~8-10m). The fluid shear rate is also on a high enough order to provide near Newtonian behavior. In our model we took the dynamic viscosity of blood to be and the density of blood to be

. In the human body, there is wave propagation through the vessel walls of the aorta, but for our model we will use rigid walls. We used the values of systolic instantaneous flow-rate of blood used in the study done by Simon, Leo, Carberry and Yoganathan in 2004 to express flow-rate as function of time to use in our simulation. We analyzed the blood flow using the COMSOL Multiphysics partial differential equation software. Due time constraints, we used a symmetrical two-dimensional model that consisted of a straight vessel with radius radius and a semicircular bulge along the upper wall with

. Mesh elements were set to a maximum size of .002 so the problem would converge. In

the COMSOL model, we used an incompressible Newtonian fluid and set its density and viscosity to the values for blood described earlier. We first solved the steady state problem with a velocity equal the peak flow velocity from the systole and then used that solution to solve the transient problem. In the transient solution, we used the flow-rate expression as function of time we found as the inlet velocity and solved the problem from 0 to .34 seconds at intervals of .01 seconds. As for boundary conditions, at the inlet we used a parabolic velocity profile to simulate fully developed flow and at the outlet we imposed a 0 downstream pressure; at each of the

walls we imposed the no-slip condition. We repeated the process for four models with varying values of ( ). Below is the model with with the mesh initialized:

RESULT

Below are pictures of the model with

at various times during the flow.

Although the simulation lasts a short time, we were able to see characteristics in the flow we would expect such as the creation of vortices at the proximal and distal ends of the aneurysm and slower flow in the aneurismal sac compared to flow in the vessel.

In the figure above, the flow-rate is at about half of the peak value it will reach in the systolic phase and flow is still laminar. We can see that maximum velocities occur at the inlet of the vessel and at the ends of the aneurismal sac. Although no slip conditions and cornersmoothing has been applied to the model, it is at these sites where vorticity arises. We also notice that flow in the center of vessel is slower than in that in the two ends due to the presence of the aneurismal sac. This velocity gradient is not present in our control model we created without an aneurysm ( .

The figure above illustrates velocity field at .26 seconds into the systole, just after the peak flow-rate is reached. Although the Reynolds number is high at this point (>4000) flow still appears to be turbulent. If our model had involved the flow due to several heartbeats as opposed to just one systole, we might have seen that flow eventually becomes turbulent as predicted by other CFD analyses. Also, in the model blood velocity has increased significantly throughout most of the vessel, yet fluid in the sac is still relatively motionless which can allow for platelet accumulation and make the aneurysm a site where thrombosis is likely to occur.

This last figure is of the velocity profile though the model at .32 seconds, right before the end of the systole. At this point the vortex that formed at the proximal end of the aneurysm has pinched off and will later make its way farther into the aneurismal sac. At this point in time, the flow-rate is decreasing with time and will actually retrograde through the diastolic phase. After this heartbeat ends, flow will then proceed in the forward direction creating more vortices, pushing them into the aneurysm creating more complex flow behavior. For each model we also used COMSOL to solve for the total pressure along the aneurysm wall versus time. We found that as increases, the peak pressure along the sac wall

increases which is consistent with Laplaces Law. As mentioned before, an increase in wall stress would eventually lead to thinning of the vessel wall which poses a greater risk of hemorrhage or rupture of the vessel. The following are plots of pressure along the aneurysm walls vs. time for each model. (Note that pressure is given in dimensional model) as opposed to due to this being a two-

CONCLUSION In our study, we surveyed the various works conducted by other research groups concerning flow blood flow through an abdominal aortic aneurysm. Of the 4 models we created, we observed the same general flow characteristics in each one. We observed that over time, a vorticity arises in blood flow due to the geometry of an aneurismal vessel. We also observed that the fluid within the main body of aneurysm moves much slower than fluid in the rest of the vessel. This would give rise to clotting of the blood due to slow moving platelets in the aneurismal sac. We also showed that as the size of aneurysm increases, so does the pressure along its walls. This pressure increase, in accordance with Laplaces Law leads to greater tension in the vessel walls which then results in vessel wall thinning. If an aneurysm is left to evolve in this manner without intervention, rupture is imminent and likely fatal. However, aneurysms have been known to rupture at small diameters in individuals so size alone is not the only measurement of an aneurysm that needs to be considered. The ideal course of action to take in the treatment of aneurysm would probably the fabrication of patient-specific models such as the one created by Wolters et al. mentioned in the literature review. However, due to computational resources and time constraints, our model didnt contain very important features concerning aneurysm geometry and behavior. Our model used rigid walls that were stationary in time and the geometry of our aneurismal sac was symmetric. As mentioned earlier in the Scotti et al. study, asymmetry is a large factor on aneurysm growth and fluid behavior. Our model also only modeled flow through the aneurysm for a very short time. Perhaps had we done a model that illustrated blood flow through an aneurysm for a longer period of time, we might have seen some more of the interesting fluid behaviors as seen in those mentioned in the literature review. Overall, with the amount computational resources available and considering the information attained by the studies mentioned in the literature review, the nature and evolution abdominal aortic aneurysms as well as the characteristics of blood flow within them are capable of being thoroughly described.

BIBLIOGRAPHY Khader, Zubair, Pai, Rao, Kamath. A Comparative Study of Transient Flow through Cerebral Aneurysms using CFD. World Academy of Science, Engineering and Technology 60 (2009): 606-610 Scotti CM, Shkolnik AD, Muluk SC, Finol EA. Fluid-Structure Interaction in Abdominal Aortic Aneurysms: Effects of Asymmetry and Wall Thickness. BioMedical Engineering OnLine (2005), 4:64 Wolters, Rutten, Schurink, Kose, de Hart, van de Vosse. A Patient-Specific Computational Model of FluidStructure Interaction in Abdominal Aortic Aneurysms. Medical 27 (2005): 871-883 Engineering and Physics Medical

Byun HS, Rhee K. CFD Modeling of Blood Flow following Coil Embolization of Aneurysms. and Engineering Physics 26 (2004): 755-761 Yip TH, Yu. Cyclic Transition to Turbulence in Rigid Abdominal Aortic Aneurysm Models. Fluid Dynamics Research 29 (2001): 81-113 Chatziprodromou AT, Tricoli A, Poulikakos D, Ventikos Y. Haemodynamics and Wall Remodelling of a Growing Cerebral Aneurysm: A Computational Model. Jourbal of 412-426 Biomechanics 40 (2007): to

Humphrey JD. Coupling Haemodynamics with Vascular Wall Mechanics and Mechanobiology Understand Intracranial Aneurysms. International Journal of Computational Fluid Dynamics, 23 (2009): 569-581 Scotti CM, Jimenez J, Muluk SC, Finol EA. Wall Stress and Flow Dynamics in Abdominal Aortic Aneurysms: Finite Element Analysis vs. FluidStructure Interaction. Computer Methods in Biomechanics and Biomedical Engineering, 11 (2008): 301-322

Hoi Y, Meng H, Woodward SH, Bendok BR, Hanel RA, Guterman RL, Hopkins LN. Effects of Arterial Geometry on Aneurysm Growth: Three-Dimensional Computational Fluid Dynamics Study. J Neurosurg 101 (2004): 676-681

Lasheras JC. The Biomechanics of Arterial Aneurysms. Annual Review of Fluid Mechanics 39 (2007): 293-319

Simon, Leo, Carberry, Yoganathan. Comparison of the Hinge Flow Fields of Two Bileaflet Mechanical Heart Valves under Aortic and Mitral Conditions Annals of Biomedical No. 12, (2004): 1607-1617 Engineering, Vol. 32,

Peattie, Riehle, Bluth Pulsatile Flow in Fusiform Modelsof Abdominal Aortic Aneurysms:Flow Fields, Velocity Patterns and Flow-Induced Wall Stresses ASME, Vol. 126 (2004): 1-9 Arroyo, L.H. & Lee R.T. Mechanisms of plaque rupture: mechanical and biologic interactions Cardiovascular Research Vol. 41 (1999): 369-375