Stress Analysis of the Aortic Valve Using Finite Element Modelling Software Toumar A.J. and Pang S.D.

Engineering Science Programme, National University of Singapore Kent Ridge Road, Singapore

ABSTRACT This project aims to create a finite element model of the healthy aortic valve in the human heart, and analyse the stresses on the valve tissue during one part of the cardiac cycle. The aortic valve is modelled geometrically and analysis is performed in the finite element software package ABAQUS/CAE. Changes in various parameters that could be associated with real life valve or circulation problems are then examined with respect to their influence on the maximum stresses of the valve tissue. The positioning of the maximum stresses is also examined and conclusions are drawn from the results of the analysis. Recommendations are made towards further investigation and model refinement. The results of this analysis can be used to identify areas of the aortic valve under greatest stress, which may be more vulnerable to rupture or damage in the case of trauma or disease than other areas. The results of the parameter analyses may provide substantial insight into how stresses are influenced by the deviation from common properties of healthy aortic valves. Thus, the study hopes to provide vital information for the prediction of possible problems with the aortic valve, and to aid in the development and implementation of prosthetic heart valves. INTRODUCTION Many studies have been made into the structure and function of the valves found in the human heart. Such studies seek to understand with greater accuracy the properties of the valves, and how these can be affected by various factors leading to valvular heart disease. The results of such studies are paramount in the design of more efficient, better functioning, economical replacement valves, which are longer lasting and safer for the individual. The aortic valve has been of particular interest with researchers, as it connects the left ventricle to the aorta, bringing oxygenated blood from the heart to the bodys largest artery, from which it is delivered to the rest of the body. The healthy aortic valve prevents regurgitation into the left ventricle, maintaining the efficiency of blood transfer. Normal wear and tear on this valve, as well as irregularities such as scars or deposits on the valve tissue, impair its function and the efficiency with which blood can be transferred through it. Calcium deposits or scarring of the valve can lead to aortic valve stenosis, which increases the workload of the heart and thus leads to a risk of heart failure (Badash, 2007). Such problems are generally treated by surgical implantation of a ball or disc prosthetic valve, bioprosthesis or an animal (porcine) aortic valve (Arcidiacono, Corvi, Severi, 2004). Manmade plastic prostheses and those made of biomaterials must therefore be designed to imitate and maintain the function of the healthy aortic valve. Techniques varying from magnetic resonance imaging (Vesely, Eickmeier, Rutt, & Campbell, 1991) to echocardiography (Shim et al., 2007) are used to give information about the geometry and material properties of the valve and its surrounding region in order to derive information for better prosthesis design. Other information, such as the influence of changes to various material parameters on the stresses within the valve, is less straightforward to obtain, though also very important. For example, the calcification of prostheses causes increased

rigidity and impaired valve function, and occurs commonly at areas of highest stress and strain (Knierbein, Rosarius, Unger, Reul, & Rau, 1992). Therefore, it is important to examine the regions of the valve in which maximum stresses occur and what factors or parameters may influence the magnitude of such stresses. Such studies can be done with relative ease and efficiency by utilising finite element software. Scope There are many factors concerning the blood and valve itself, which may influence parameters within a computer model and also the magnitude of stresses on the valve tissue. Deposits on the valve walls, for example, may be approximated by a change in the coefficient of friction between the blood and valve interior. Calcification and scarring changes the way the valve can be stretched, which can be simulated by changes in the elasticity properties of the computer model. Blood consistency may change due to a disease of the blood, or the use of artificial blood in blood transfusions (Winslow, 2006). This would in turn change properties such as the viscosity or shear modulus of the finite element blood model. Changing these parameters within a controlled computer model requires much less effort and time expenditure than observing their effects in nature, and is less invasive than removing and examining animal heart valves. In addition, a computer model allows just one parameter to be changed at a time, thus avoiding confounding variables that would more likely be present in in vivo observations. This study examines the effect of changes in valve elasticity and blood viscosity on the stresses in the aortic valve tissue. Once the results of this analysis on the open valve are verified, this model can be coupled with that of the closed valve in order to simulate an entire cardiac cycle. As well as aiding in understanding of what influences the stresses on the heart valve, the results of this study may lead to the improvement of prosthesis designs. MODELLING Geometry of the Aortic Valve As stated previously, the aortic valve joins the aorta to the left ventricle. It is asymmetric, and consists of three leaflets and three sinuses (cavities behind the leaflets). The sinuses can be viewed as approximately ellipsoid dilations of the aorta, to which leaflets are connected at the base of the valve (Thubrikar, 1990). During the cardiac cycle, the valve leaflets open and are pushed out into the sinuses by the flow of blood. It is in this position that the leaflets were modelled in this investigation, according to the modelling method suggested by Thubrikar and dimensions typical of a healthy human aortic valve (Gnyaneshwar, Kumar, & Balakrishnan, 2002). The finite element software package ABAQUS/CAE was used to create the geometric model and to run the analysis. The three leaflets of the aortic valve were modelled with their region of attachment to the base of the aorta. The aortic valve was modelled in the fully open position, with the three leaflets in the fully expanded position within the sinuses (not included in the model). Table 1. Dimensions of Aortic Valve Model Used in Investigation
Dimensions of Aortic Valve Model Radius of Commissures Radius of Aortic Base Height of Valve Height of Commissures Angle of free edge of leaflet to plane passing through commissures Angle of undersurface of leaflet to plane passing through commissures 12.0 mm 12.0 mm 17.0 mm 8.52 mm 32 degrees 22 degrees

Modelling of the Open Aortic Valve The three leaflets of the aortic valve were modelled as cylindrical shell elements stemming from a circular shell base. The leaflet and base tissues were modelled initially with an isotropic, linear elastic material with a Youngs Modulus of 2.0 MPa (Ranga, Mongrain, Biadilah, & Cartier, 2007) and a Poissons ratio of 0.3 (Gnyaneshwar et al., 2002). The thickness of the leaflets in the human aortic valve was found to vary between approximately 0.25 and 1.33 mm throughout each leaflet, however in this model, a uniform thickness of 1 mm was used (Thubrikar, 1990). Further modelling and analysis could take into account the variation in thickness of the valve elements, however for this model, a simplified representation was created. The valve material model would be improved by the use of a hyperelastic material, instead of an elastic material, however for the elastic model created, parameters were chosen within the measures of healthy human aortic valves in order to simulate the valve tissue as accurately as possible.
Valve Model Commissural height 8.52 mm Valve height 17 0 mm 4 mm Base Radius 12.0 mm Blood Model

Figure 1: Features of the Blood and Open Valve Model in Viewport of ABAQUS/CAE Blood flow was modelled by the procession of a 4 mm deep section of blood through the open valve model. This section was modelled to match the internal geometry of the valve model at the base of the leaflets, so that contact could be established at the initial incremental time-step of its motion. As the modelling and analysis were conducted in ABAQUC/CAE, the fluid blood section was modelled as a Neo-Hookean hyperelastic solid with a low initial shear modulus. Further studies can improve on the model by utilising fluid and solid interaction software, and more accurately depicting the viscosity and consistency of blood flowing from the heart. Contact was established between the blood section and interior walls of the valve and leaflets. Several coefficients of friction between the blood section and leaflets were tested against the maximum Mises equivalent stresses on the valve tissue. Significant changes in the friction coefficient were found to have little effect on the magnitude of the stresses within the model, with maximum stresses varying as little as 0.1% for doubling the reference friction coefficient; hence an arbitrary, small coefficient of friction was used for all subsequent analyses. These analyses were used to test the effects of changing the elasticity of the valve, and the blood consistency, on the Mises equivalent stresses in the valve. Meshing The solid part representing a section of blood fluid was meshed with 10-node modified tetrahedral (C3D10MH) elements, while the homogeneous shell structure was meshed with

triangular, general-purpose shell (S3) elements. For initial studies, a small number of elements were used for both parts due to restrictions of the student edition of ABAQUS 6.7. ANALYSIS OF RESULTS The regions of maximum stress were examined visually throughout the blood motion. The analysis was then repeated to test the influence of a change in the valve elasticity and blood viscosity against the reference model of a healthy, open human aortic valve. Position of Maximum Stresses The movement of the blood section lead to a range of stresses, apparent on the valve model shown in Figure 2 below. Maximum Mises equivalent stresses were found to be at the attachment of the leaflets to the base of the aorta (highlighted in the top left image), as can be seen from these images of four stages of the blood motion.

Figure 2: Maximum stresses on the valve model (shown circled) in four stages of blood motion. Maximum Mises equivalent stresses were apparent throughout the analysis at the regions of attachment of the leaflets to the base of the aorta. These hinge regions would thereby need to withstand the highest stresses during blood flow, and must be reinforced during the design for prostheses. This positioning of the maximum stresses on the valve tissue is consistent with the thickness of the leaflets of natural valves, which tend to become thicker and thus more resistant to tearing at their bases (Thubrikar, 1990). Valve Elasticity In natural as well as prosthetic valves, calcium deposits on the valves or changes in the tissue associated with aging or scarring around the valve (Kuehn et al., 2008) influence the elasticity of

the leaflets and sinuses, making them more stiff and rigid. It is known that such changes may impair the functionality of the valve and demand extra pressure from the heart, increasing the risk of heart failure in the long term. It is not as widely investigated what such a change in elasticity does to the mechanical stresses within the valve itself. In theory, an increase in the magnitude of stresses on the valve would increase the risk of further damage, and this factor was investigated in this study. The valve model was tested with an increasing Youngs modulus, and the stresses caused by the blood flow in the stiffer valve were compared to the reference values. The Mises equivalent stresses were evaluated within an incremental analysis step with 100 increments, and data from three representative increments (coinciding with 3 positions of the blood section within the valve) were recorded. The stages chosen were at Increment 40, 60 and 80 to avoid integration problems that may have occurred at the beginning and end of analysis. Figure 2 shows the plot of the normalised stress data against the relative increase in Youngs Modulus. Table 2: Normalised Maximum Stresses at Increments 40, 60 and 80
Normalised Normalised Maximum Stress at Youngs Modulus Increment 40 Increment 60 Increment 80 1 1.000 1.000 1.000 1.25 1.162 1.128 1.091 1.5 1.311 1.238 1.166 1.75 1.449 1.336 1.229 2 1.579 1.424 1.283 2.25 1.700 1.503 1.331 2.5 1.814 1.576 1.353 2.75 1.921 1.643 1.394 3 2.023 1.704 1.431 3.25 2.119 1.762 1.464

Increment 40

Increment 60

Increment 80

Change in Valve Elasticity versus Maximum Mises Equivalent Stress

2.500E+00 2.000E+00 1.500E+00 1.000E+00 5.000E-01 0.000E+00 0 1 2 3 4 Norm alised Young's Modulus (MPa/MPa) Increment 40 Increment 60 Increment 80

Figure 3: Maximum Mises Equivalent Stress vs. Change in Valve Elasticity at Various Increments of Analysis

Normalised Maximum Stress (MPa/MPa)

An increase in stiffness of the leaflets and aortic root lead to an increase in the magnitude of stresses measured during contact between the blood section and the valve. Since higher stress areas tend to accumulate calcium deposits, and thus cause further stiffening in the valve, it is important that prostheses are designed to mimic the elastic flexibility of natural, healthy aortic valves. The stresses increase most prominently at Increment 40, which corresponds to the lower part of the valve model, where leaflets are attached to the aortic base. The increase in stresses appears to become less prominent as higher Youngs Moduli are tested. This is indicative of a non-linear relationship between the stress and Youngs Modulus of the tissue. Stress and Youngs Modulus in an isotropic object can be related thus:

stress strain

The trends observed are fitting with this equation, as the stress is proportional to both the increased Youngs Modulus and the decreasing strain. The three differing trends observed at the three positions in the valve model may be attributed to the differing geometry throughout the valve, as the leaflets (at the top of the valve shown in Table 2) are free to move while the base of the aortic valve is fixed with displacement constraints. Blood Consistency Diseases in the blood or the use of blood substitutes in patients influences the viscosity of the fluid passing through the aortic valve during the cardiac cycle (Winslow, 2006). The change in viscosity and its effect on the stress in the aortic valve were investigated in this study. The section of blood flowing through the valve model was modelled as a Neo Hookean hyperelastic solid. An increase in blood viscosity was therefore approximated by increasing the value of the initial shear modulus (0 = 2C10), by changing the temperature dependent parameter C10 in the material model (Dassault Systmes, 2007). Stresses caused by this more viscous blood material were compared with reference values at three increments of the blood motion step. Figure 4 shows the plot of the normalised stress data against the relative increase in C10. Table 3: Normalised Maximum Stresses at Increments 40, 60 and 80
Normalised Normalised Maximum Stress at Material Parameter C10 Increment 40 Increment 60 Increment 80 0.5 8.068E-01 7.339E-01 6.537E-01 0.75 9.195E-01 8.880E-01 8.555E-01 1 1.000 1.000 1.000 1.25 1.063 1.086 1.116 1.5 1.113 1.154 1.211 1.75 1.156 1.210 1.291 2 1.198 1.328 1.359 2.25 1.225 1.379 1.418 2.5 1.254 1.423 1.470 2.75 1.279 1.462 1.516 3 1.380 1.524 1.558 3.5 1.615 1.759 1.630 4 1.829 1.994 1.689

Increment 40

Increment 60

Increment 80

Change in Initial Shear Modulus (C10) versus Maximum Mises Equivalent Stress
2.500E+00 2.000E+00 1.500E+00 Increment 40 1.000E+00 5.000E-01 0.000E+00 0 1 2 3 4 5
Normalised Initial Shear Modulus (MPa/MPa)

Normalised Maximum Stress (MPa/MPa)

Increment 60 Increment 80

Figure 4: Maximum Mises Equivalent Stress vs. Change in Initial Shear Modulus at Various Increments of Analysis Predictably, an increase in the viscosity or solidity of the blood passing through the heart valve increases the magnitude of stresses in the valve itself. The blood shears less readily, exerting a higher force on the valve. For all three stages of blood flow examined, the relationship between the initial shear modulus of the blood and the maximum Mises equivalent stress in the valve tissue appears to be non-linear. At increments 40 and 60, the relationship between increase in C10 and increase in stress seen in Figure 4 appears to be consistent with that at increment 80, but this tends to break down at higher values of C10. The initial shape of the curves suggests a logarithmic relationship, after which the values at 40 and 60 appear to become linear. This inconsistency may be due to the particular geometry of the valve itself. As can be seen in Table 3 above, the open valve is significantly wider at the free edge than at its base. Inconsistency in results at different increments may also be due to inadequately refined mesh, which was limited by the use of the student version of the software. The more powerful, professional version should be used in future investigations to refine the mesh. Since blood substitutes are used in blood transfusions, which may be necessary during surgery (Winslow, 2006), it is important that their viscosity be monitored and maintained at a level close to that of natural blood, to decrease the stress on the heart valves, which may lead to extra demand on the heart itself Recommendations for Future Investigations There are many ways in which this simulation can be expanded and improved in order to achieve more plausible and useful results. Several aspects of the geometry of the model can be refined, such as the variation in thickness within the valve base and leaflets, the addition of sinuses and muscular reinforcements surrounding the valve and the slight asymmetry of the three leaflets observed in nature. The geometry of the closed and opening valve can be included in the model in order to widen the scope of the model to that of the entire cardiac cycle, in order to analyse the stresses on the valve leaflets during the opening or closing of the valve, where they may suffer the largest mechanical stresses. Another area of improvement is the material model of the valve tissue. An isotropic, linear elastic material model is not entirely reflective of the valve tissues anisotropic, hyperelastic

nature. A more accurate model would be one that would deform nonlinearly a hyperelastic model. This would require additional data in order to model in ABAQUS, including biaxial tension data, which may be found in literature or through experiment (Gnyaneshwar et al, 2002). It is recommended that further investigations use a more refined mesh, in order to utilise the valve geometry more accurately during analysis. CONCLUSIONS The development of computer software has allowed for the nature and effectiveness of biological systems and their prostheses to be modelled and investigated without the use of invasive or time consuming observational techniques. The hearts four main experience changing pressure, stresses and wear continually, and this, together with forms of valvular or blood disease, scar tissue from previous surgeries or calcium deposits in higher stress areas, can change the material properties of the valve tissue. The aortic valve, connecting the heart to the bodys largest artery, is of particular importance and the subject of much study. The effects of wear and tear and other influences discussed in this study are difficult to calibrate and observe in nature, but can be better understood and approximated with relative ease when modelled and analysed appropriately using finite element software. These modelling techniques allow for a greater range of conceptual testing for prostheses, and allow material and shape properties of prosthetic aortic valves to be optimised in the design stage, leading to faster evolving prosthesis designs with improved functionality and thereby an improved outlook for the patients who require them. REFERENCE Arcidiacono, G., Corvi, A., Severi, T. (2004). Functional analysis of bioprosthetic heart valves. Jbiomech, 38, 1483-1490 Badash, M. (2007) Aortic Stenosis. Retrieved from http://www.mountsinai.org/Other/Diseases/Aortic%20stenosis Dassault Systmes (2007). Abaqus analysis users manual. Gnyaneshwar, R., Kumar R. K., & Balakrishnan, K. R. (2002). Dynamic analysis of the aortic valve using a finite element model. The Annals of Thoracic Surgery, 73, 1122-1129. Knierbein B, Rosarius N, Unger A, Reul H, & Rau G. (1992). CAD-design, stress analysis and in vitro evaluation of three leaflet blood-pump valves. Journal of Biomedical Engineering, 14(4), 275-86. Kuehn, A., Baumgartner, D., Baumgartner, C., Hoerer, J., Schreiber, Ch. Hess, J. et al. (2008). Impaired elastic properties of the ascending aorta persist within the first 3 years after neonatal coarctation repair. Pediatr Cardiol, 30, 4651 Ranga, A., Mongrain, R., Biadilah, Y., & Cartier, R. (2007). A compliant dynamic FEA model of the aortic valve. 12th IFToMM World Congress. Shim, C. Y., Watanabe, N., Tsukiji, M., Yamaura, Y., Ogasawara, Y., Ha, J.W. et al. (2007) Three-dimensional geometry of aortic valve: A new trial of visualization with real-time threedimensional echocardiography. Journal of Echocardiography, 5(2), 55-57.

Thubrikar, M. (1990). The aortic valve. USA: CRC Press. Vesely, I., Eickmeier, W., Rutt, B., & Campbell, G. (1991). Analysis of the aortic valve geometry using dynamic, three-dimensional display. Engineering in Medicine and Biology Society, Proceedings of the Annual International Conference of the IEEE, 13, 1181-1182. (MRI) Winslow, R.M. (Ed.). (2006). Blood substitutes. London: Academic Press.