A fundamental footprint of an impeller design in an artificial heart pump has remained unchanged since the first design of rotary ventricular assist device (VAD) in the 1980s which replaced pulsatile designs.1 Multiple blades attached to the central hub with varying angles, lengths, thickness, gap sizes, etc., have been extensively explored to optimize hemodynamic performance while minimizing hemolysis and thrombosis for ventricular assistance.2,3 Significant effort has been made to reduce mechanical components contacting blood while retaining the framework of impeller throughout design iterations.4 Despite the improvements in survival and quality of life achieved with advances in VAD technology, adverse events such as strokes, thrombosis, and gastrointestinal bleeding remain a significant problem.5,6 Percutaneous VAD has received increasing attention since it reduces invasiveness compared with current VAD therapies, leading to an overall improved quality of life. However, hemolysis is still a major problem that results in short-term use.7–9 The tradeoff of small footprint with minimal/no friction (durability) is a large central impeller forcing blood through narrow side passages.
In a search for an ideal impeller design, we extensively studied comparative anatomy/physiology across the phyla. Anatomically mammalian heart and avian heart appear similar in terms of four-chamber architecture. However, avian species evolved from therapod dinosaurs about 231 million years with a distinct evolutionary advantage over mammalian evolution with the avian right ventricle being more advanced. Birds not only have to fly in extreme environments but also have to change their pulmonary output based on the resistance at high and low altitude. From rest to flight, they have to quickly adjust the pressures as they go against gravity and come down with gravity and accommodate significantly stressful conditions while migrating at high altitude. Thus, the avian right ventricle is physiologically far more adapted to changes in output based on flow and resistance, with lower heart rates per unit size for greater perfusion of the body.10,11
A conspicuous absence of the one-way valve is a key element of the avian right atrioventricular valve. Instead of the one-way valve, the avian right-sided atrial ventricular wall is replaced by a spiral flap (Figure 1A). As this spiral flap twists and contracts, its morphologic changes including the pitch, length, height, etc., regulate the flow and pressure. However, it is impractical to use a single spiral flap with the rotary pump principle because a rotation of asymmetrically attached single spiral blade would not create a forward flow. We investigated in considerable details including an optimal diameter, helix height, helix pitch, blade length, blade thickness, blade angles, number of blades, and blade shape needed to have a pumping mechanism actuated by a rotary pump principle.
We present the initial characterization and hydraulic/hemodynamic performance of a biologically inspired, open, helicoid (BiO-H) impeller.
The clay modeling shown in Figure 1B describes our initial modeling process of BiO-H inspired by the avian heart valve. We started with attaching a single spiral blade inside the bushing wall just like a single flap attached to the atrial ventricular wall in birds (Figure 1A). However, a rotation of asymmetrically attached single spiral blade would not create a forward flow. To properly combine this new impeller design with the rotary pump principle for the actuation, we came up with three angled blades spirally attached to the cylinder wall with an unobstructed bore at the center (Figure 2A). To optimize the design in terms of pump performance (e.g. volumetric flow rate, head pressure, torque), eight influential parameters have been defined, inner diameter (d), helix height (h), helix pitch (p), blade length (ℓ), blade thickness (t) (mm), introductory blade angle (α), number of blades (n), and blade shape (Figure 2B).
Computational Fluid Dynamics
Initial optimization studies for the BiO-H impeller design were performed through the SolidWorks Flow Simulation module using the setup described in Table 1. The assumed actuation was considered to be in an axial format with the impeller suspended within a ring electromagnet. Parametric analysis of geometrical variations on the BiO-H impeller design aimed at maximizing volumetric flow rate along with assessing head pressures and torque. Two final models, Mod 1 (Figure 3A) and Mod 2 (Figure 3B), determined based on these initial optimization studies were examined in detail with Ansys software (Ansys Inc., Canonsburg, PA). Mod 2 is a refined model based on Mod 1 simulation results to see how the simple design modifications (Table 2) affect overall performance. In Mod 2, tapered ends for a smooth blood transition into the gap were implemented. The cutoff at the leading edge in Mod 1 was replaced with the rounding edge to decrease hydraulic losses. The trailing edge was transformed into a vertical cutoff. Among all, the key features added to Mod 2 were the troughs on the outside with the same wrap angle as that of impeller blades, which is expected to prevent the backflow through the outer gap and, thus, increase overall pump efficiency.
Table 1. -
Simulation Setup in SolidWorks Flow Simulation Module for Parametric Study and Initial Design Optimization
||310 K (Body temperature)
||Adiabatic, no roughness
|Inlet boundary condition
||60 mm Hg
|Outlet boundary condition
||120 mm Hg
||120 mm Hg
Table 2. -
Design Parameters for Mod 1 and Mod 2
The geometrical setup allowed controlling mesh regions from 11 to 17 for adequate control of the mesh parameters. The simulation setup for our impeller design was composed of seven domains: inlet, inlet pipe, outlet, outlet pipe, impeller, center bore, and radial gap (Figure 3C). For better control of the meshing, the domains were further divided into more mesh regions if necessary. An elongated inlet and outlet domain of 80 mm was implemented to decouple the pump hydraulic prediction from any flow effects, e.g., vortex formations, near at the inlet and outlet. For Mod 1 and Mod 2, there were 12 and 17 mesh regions with overall mesh consist of 4,418,397 elements and 13,299,271 elements, respectively. The meshing of Mod 2 required more mesh regions to ensure a sufficient mesh quality in the gap, the additional tapered ends, the troughs on the outside, and the roundings at the leading edge; these additional features resulted in larger elements. Impeller domains were mainly meshed in structured hexahedral meshes using Turbogrid (Ansys Inc., PA). All other domains were meshed using Meshing (Ansys Inc, PA). Prism layers and gap refinements were used to resolve wall shear adequately in all domains. The shear stress transport (SST) k-ω turbulence model with a residual target of 1e-4 and a total of 250 iterations per operation point was used with a varying time step (e.g., 40 × 0.01 s, 40 × 0.001 s, 20 × 0.0005 s, 150 × 0.0001 s) to allow transient response to become settled. Steady-state simulations were performed using an alternate rotating model with “frozen rotor” as frame change model. A mass flow boundary condition for inlet combined with an opening pressure boundary condition for outlet promised a suitable setup for a stable steady-state simulation and valid results. The pressure head, gap, bore flow, and impeller flow were monitored during the solver run to assure convergence. No mesh independence was considered necessary. The mesh size has been chosen based on grid sizes in respective studies with a similar focus of investigating hemodynamic performance in rotary blood pumps (Different axial flow VADs: 1.9–5.2 million elements12; CentriMag: 4.86 million elements, HeartWare Ventricular Assist Device (HVAD): 15.7 million elements, HeartMate II: 3.5 million elements in a third periodic mesh13; CentriMag: 3 million elements14; Different VAD designs: 4.7–7.6 million elements3) (see Figure S1, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467). Simulations were performed at three different pump speeds: 10,000, 12,500, and 15,000 revolutions per minute (RPM). Three rotating speeds chosen in the current study are within the operational speed range of the existing device which is from 6,000 to 15,000 RPM. To save time and to quickly look at the impact of the additional features, we evaluated three data points for Mod 2: 5, 10, and 20 L/min at 15,000 RPM. Blood, a non-Newtonian fluid, was used as the working fluid with the blood shear thinning properties according to the model presented by Ballyk et al.15 The simulations for Mod 1 took 6.316e+4 central processing unit (CPU) seconds for each design point. A single solver run for Mod 2 took approximately 1.177e+5 CPU seconds.
Mod 1 has been modified to actuate with a shaft-driven mechanism. We made three-dimensional printed prototypes with two different impeller diameters, 8 mm and 15 mm, in which the larger design corresponds to the one used in the computational fluid dynamics (CFD) study. For 8 mm impeller, we uniformly scale down 15 mm model using the scale function in Autodesk Inventor (Autodesk Inc., San Rafael, CA). Whereas the 15 mm impeller was made of steel, the 8 mm impeller was made of aluminum due to the technical challenge of three-dimensional printing thin wall with certain materials (e.g., steel). We used a commercial three-dimensional-printing service, Shapeways, for printing the metal impellers. A process called Selective Laser Melting was used for aluminum (alloy AlSi10Mg, 10% Silicon 0.5% Mg) three-dimensional prints with a resolution of 25 µm. For steel three-dimensional prints, a binder jetting process which is known to have a resolution around 80 µm was used and was later infused with bronze, creating a material that is 60% steel and 40% bronze.
A tripod configuration with the circular column was incorporated at one side of the impeller and directly connected to the motor shaft by the mechanical coupling (S50FP9MFB153008, SDP-SI, Designatronics Inc., Hicksville, NY) (Figure 4A). The three-dimensional printed BiO-H impeller coupled with brushless micro-DC motor (see Table 3 for motor specifications) was then encased in a three-dimensional printed housing that has 3/8-inch tubing connector configurations for inlet and outlet (Figure 4B). Pump housings were printed using Objet30 Pro three-dimensional printer having a resolution of 28 µm with proprietary resin that is acrylonitrile butadiene styrene like.
Table 3. -
|Brushless Micro-DC Motor
||30 × 12, 38 (with shaft)
|Diameter of shaft (mm)
|Length of front shaft (mm)
|Lamination thickness (mm)
||2 Cell 2S/7.4
|Maximum current (A)
|No load current () (V/A)
|Current capacity (A/s)
|Internal resistance () (Ω)
Benchtop Experimental Setup and Measurements
A mock circulation loop equipped with an ultrasonic flow sensor (ME8PXL; Transonic Systems, Inc., NY) connected to a flow meter (TS410; Transonic Systems, Inc., NY), pressure transducers (MLT0380; AD Instruments, Dunedin, New Zealand), and clamp was used to create a pump performance curve for each prototype (Figure 4C). The prototype was immersed in water and run with a motor controller (Lantian RC 6-40V Multifunctional Motor ESC Propeller Tester for RC Drone), which provides a user interface to control the motor speed by either a knob or an embedded algorithm that quantitatively controls a throttle, for example, by the increment of 10% and records data including voltage, current, and power. Loaded RPM was estimated by following the equation shown below:
where motor Kv, , and are motor constants shown in Table 3, η is the efficiency defined by the ratio between input () and output power (), throttle is the amount of power provided by percentage in the motor controller based on the pulse width modulation technique, and and are voltage and current measured at motor terminals, respectively.
Computational Fluid Dynamics
Figure 5 summarizes the results of the parametric analysis. Volumetric flow rate increases with a larger diameter. However, a larger diameter comes at the price of a larger footprint. Volumetric flow rate increases with height up to a helix height of approximately 20 mm. Increases in height above 20 mm lead to very minimal changes in volumetric flow rate, allowing the optimal height to be approximately 15–25 mm. The optimal pitch to maximize volumetric flow rate varies with the helix height. Torque decreases with an increase in pitch. Volumetric flow rate increases with a blade length up to a maximum length of d/2, where d is an inner diameter, whereas head pressure and torque increase with the blade length. Volumetric flow rate slightly decreases with an increase in blade thickness, but the effect is moderate. Blade thickness has a minimal effect on head pressure and torque as well. Blade shape also has a minimal effect on the volumetric flow rate. Through this initial design optimization in respect to each parameter, final models, Mod 1 and further refined model, Mod 2, based on Mod 1 simulation results have been created for detailed analysis in regard to pump performance, efficiency, and flow field (Figure 3).
A numerically determined pressure head at flow rates and at three different pump speeds (10,000, 12,500, and 15,000 RPM) gives the first insight into the pump performance of Mod 1 (Figure 6A). Pump performance curve of Mod 1 shows that an adequate flow–pressure relation for a right ventricular assist device operating at full support (>5 L/min at 30-40 mm Hg)16 can be achieved at a pump speed below 10,000 RPM and an adequate flow–pressure relation for a left ventricular assist device operating at full support (>5 L/min at 80–120 mm Hg) at a pump speed of 15,000 RPM. Similar to common blood pumps, the pump hydraulic efficiency over different flow rates at the three different pump speeds shows a tendency to have higher efficiency for higher pump speed (Figure 6B).
Flow field analysis has been carried out at the best efficiency point (BEP) for each RPM. For example, BEP for 15,000 RPM is at the flow rate of 20 L/min (VBEP = 20 L/min) according to Figure 6B. Here, hydraulic power which visualizes efficiency of the flow guidance and highlights areas where hydraulic losses occur has been evaluated and shown as a variable on the impeller (Figure 7). The integral of this variable over the full impeller surface yields the shaft power as an absolute value (see Figure S2, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467). For all three pump speeds, negligible hydraulic losses were found in the gap between bushing and housing compared with those on the blades (Figure 7, A and C). In a single blade analysis evaluated at pump flows below (V < VBEP), at (V = VBEP), and above (V > VBEP) the BEP, a hot spot of the hydraulic power is found on the top surface, right behind the leading edge, typically caused by flow separation (Figure 7, B and D). This is usually the case when the blade angle of the leading edge is too high. This assumption is proven by the fact that this spot decreases with the flow but increases with the pump speed (see Figure S2, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467). The inner half of the blade does not provide any flow guidance to the flow. The horizontal cutoffs as leading edges cause flow impingements and, thus, an increased hydraulic power.
The hydraulic flow field inside the Mod 1 shown in Figure S3A, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467, exemplarily for 12,500 RPM presents backflow in the gap between bushing and housing. The backflow is linearly dependent on the pressure gradient. A certain amount of backflow is necessary to ensure the washout of the gap region, but within the range that does not negatively affect pump efficiency. The backflow is quantified in Figure S3B, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467.
Based on the Mod 1 simulation results, the second design (Mod 2) has been created. The altered pump performance curve shown in Figure 6A from that of Mod 1 demonstrates that the simple design change can provide a considerable impact on the pump hydraulics. Mod 2 created a steeper pump curve, which means pressure drop results in a smaller change in flow. In other words, the flow will be more stable against pressure variations. The hydraulic efficiency curve shows the tendency of a more left-sided skewed curvature peak, resulting in altered BEP for Mod 2 (Figure 6B, marked by the red asterisk). The shaft power was decreased but in a disproportional relation to the pressure head (see Figure S2, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467). The secondary gap backflow substantially decreased compared with Mod 1 attributed to the implementation of the troughs on the bushing (see Figure S3B, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467). The influence of the tapered ends remains disputable because the tangential velocity component in and before the gap is predominant. However, the tapered ends cause a reduction of the vortices production within the impeller inlet region. Similar to the Mod 1 results, the contour plots of the shaft power as a variable on the impeller show that the outer trough has a negligible influence on the overall integrated shaft power compared with the blade passages (Figure 7B).
In Vitro BiO-H Impeller Testing
Figure 8, A and B shows pump performance curves for small and large BiO-H impeller, respectively. As expected from the CFD analysis (Figure 5), large BiO-H impeller resulted in greater volumetric flow rate. Throttle, the amount of power provided by percentage in the motor controller, was increased by 10% every 5 s from 10% to 100%. Current and voltage measurements at each increment have been recorded to calculate power efficiency and estimate the loaded RPMs. The calculated power efficiency (see Figure S4, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467) shows a similar trend to the simulation result shown in Figure 6B. Although the power efficiency was higher for the large BiO-H impeller until it reaches the BEP, it dropped more rapidly and the maximum RPM results in lower value for the large BiO-H impeller (17,000 RPM) than the small BiO-H impeller (20,000 RPM), which implies that higher load has been applied to the motor coupled with the large BiO-H impeller than the one coupled with the small BiO-H impeller. Not only the size but also the material used for the large BiO-H impeller, in which steel is much heavier than aluminum, attributed to the higher load. There were only minimal differences in performance between 90% and 100% throttle probably due to power limit that was reached. Hence, performance curves at 90% throttle have been excluded for both impellers. The estimated RPMs were rounded to the nearest 1,000 (Figure 8).
Compared with the simulation results, the benchtop experiment with three-dimensional printed prototype showed a better pressure–flow relationship, higher pressure head at the same flow rate. For example, the maximum pressure head reached at 5 L/min was ≈220 mm Hg in the benchtop experiment, while ≈130 mm Hg in the simulation. The small BiO-H impeller reached a maximum flow rate of 6.2 L/min against 63 mm Hg at 22,000 RPM, and the large BiO-H impeller reached a maximum flow rate of 8.3 L/min against 120 mm Hg at 17,000 RPM.
Pulsatile pumps can provide hemodynamically superior performance; however, the need for multiple moving parts makes them prone to rapid wear and tear. Rotary pumps, on the other hand, place the actuation mechanism within the path of blood and almost eliminates multiple moving parts that are responsible for friction associated mechanical failure. However, rotary pumps have smaller stroke volume and subsequently need for higher RPMs to achieve output. Exploiting the advantage of the rotary pump with a small footprint and durability but increasing the stroke volume per revolution is an ideal engineering goal to achieve.
We present a novel impeller design inspired by avian cardiac morphology, BiO-H impeller, in which spiral blades attached to the inner shell create flow without obstruction by the central impeller body. Considering remarkable pump efficiency that birds have to accommodate swimming, running, and flying, we believe structural features of the avian heart valve, which has been relatively unexplored, can provide different perspectives on impeller design.10
The main purpose of this study is to present a new impeller design that has fundamentally different footprint from the conventional one and evaluate the performance of it. We were able to show its promising results in benchtop experiments even with the raw state of the three-dimensional printed prototype as well as in CFD analysis.
In the benchtop experiment, at 10,000 RPM, BiO-H produced 5 L/min at 40 mm Hg as similar to HeartMate II (Abbott, Chicago, IL) which produces 6 L/min at 50 mm Hg.17 However, considering BiO-H is a raw prototype without any flow-assisting components included such as inducer, diffuser, or flow straightener, while HeartMate II is a market product having all those supplemental components, we can conclude that BiO-H is a potentially promising new impeller design in respect to its hydraulic performance. With flow assisting components incorporated if necessary, BiO-H can be operated at lower rotational speed with improved efficiency.
Moreover, BiO-H has a large gap size, fivefold greater than that of HeartMate 3 (Abbott, Chicago, IL). The number of blood cells that can be stacked inside the BiO-H is 838 compared with 8 and 163 for the HeartMate II or HeartWare HVAD (Medtronic, Minneapolis, MN) and HeartMate 3, respectively (see Figure S5, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467). Considering that shear stress is directly related to the gap size, such a large flow path equivalent to small- to medium-sized human arteries (6.7 mm dia.) has a promising potential to provide more physiologic flow avoiding hemolysis, although the design comes down to the percutaneous level. Another way of showing an improved flow path is by measuring the occluded cross-sectional area. BiO-H shows 2.25-fold less cross-sectional area loss than the conventional impeller design (e.g., HeartMate II) (87.37% vs. 39.75%) (see Figure S6, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467).
According to our initial optimization process, the 8 mm device, which can provide a percutaneously deployable pump design, creates a sufficient amount of flow (5 L/min at 80 mm Hg) at 15,000 RPM. Instead of spending time evaluating and optimizing the 8 mm device through simulation, we decided to jump to benchtop experiments and compare the performance between 15 and 8 mm. Data collected from the benchtop experiment can provide a useful basis for follow-up hemolytic characterization study with further design modifications and optimizations for both 8 and 15 mm impellers.
The benchtop experiment showed promising results even with the unoptimized three-dimensional printed prototypes. High precision manufacturing and assembly with enhanced surface finish, durability, and stability can significantly improve the overall performance. Hub ratio, which has been reported as one of the important determinants of the pump efficiency, can be another potential parameter for the optimization.18
Limitations and Future Work
Our current design of BiO-H pump has right angle flow change between inlet and outlet, which is similar to the existing VADs; this is attributed to the anatomical arrangement between ventricular apex and aorta. In the current study, we focused on performance evaluation of our new impeller design and demonstrated its successful pressure–flow relationship without any particular inlet and outlet configurations in consideration. However, the inlet and outlet arrangement and their duct designs are important parameters that affect hemocompatibility, as well as hemodynamic performance because their roles are to guide, straighten, equalize the incoming flow, and minimize the energy loss in the outgoing flow.19 Improper design of inlet or outlet will result in separated flow, recirculation, and energy loss, leading to mechanical blood damage.20,21 For example, trumpet duct shape significantly affects flow profile by slowing down the flow and increasing pressure while reducing the flow in accordance with the Bernoulli principle. We will investigate the ideal inlet and outlet shape and their arrangements to optimize the pump performance through simulations, as well as benchtop experiments in future work.
The follow-up hemocompatibility evaluation will include modeling von Willebrand factor cleavage, platelet activation, residence time, turbulence shear stress, the interplay between shear rates and exposure time, and recirculation ratio which will allow providing comparative analysis with other devices.22 We will address hemolytic characteristics of BiO-H in a subsequent experiment through both CFD with hemolysis prediction models and benchtop hemolysis testing following American Society of Testing and Materials (ASTM) standard.
We will also investigate the optimal pump actuation mechanism through strategic combinations of the actuation mechanisms described in Figure S7, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467. There have been two different actuation mechanisms to run the impeller in blood pumps, shaft driven and rim driven. The shaft-driven mechanism requires mechanical coupling between the motor shaft and the impeller shaft, while the rim-driven mechanism requires magnetic coupling with or without mechanical bearings (e.g., pivot, hydrodynamic, thrust, ball bearing), instead of mechanical coupling.17 Both mechanical bearing and magnetic coupling can improve the stability of the impeller axially and radially while it rotates not only in the rim-driven but also in the shaft-driven mechanism (see Figure S7, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A467).
We presented an evaluation of our novel BiO-H impeller design which has a fundamentally different footprint from the long-standing traditional Archimedean screw-based impeller design. Simulation results show the unprecedented capability of the BiO-H impeller design in generating high flow, and the benchtop experiments corroborate this promising performance by showing further improved pressure–flow relationship compared with the CFD results. Such a novel design has the potential to create more physiologic blood flow path and reduce adverse event associated with rotary pumps.
1. Nosé Y, Yoshikawa M, Murabayashi S, Takano T. Development of rotary blood pump technology: Past, present, and future. Artif Organs 2000.24: 412–420.
2. Liu L, Wang FQ, Wu QL, et al. Influence of impeller design on hemolysis of an axial blood pump. Appl Mech Mater 2012.140: 162–166.
3. Wiegmann L, Boës S, de Zélicourt D, et al. Blood pump design variations and their influence on hydraulic performance and indicators of hemocompatibility. Ann Biomed Eng 2018.46: 417–428.
4. Feldmann C, Chatterjee A, Haverich A, Schmitto JD. Left Ventricular Assist Devices–A State of the Art Review. InHeart Failure: From Research to Clinical Practice
, Springer, Cham 2018, pp. 287–294.
5. Emani S. Complications of durable left ventricular assist device therapy. Crit Care Clin 2018.34: 465–477.
6. Whitson BA, Eckman P, Kamdar F, et al. Hemolysis, pump thrombus, and neurologic events in continuous-flow left ventricular assist device recipients. Ann Thorac Surg 2014.97: 2097–2103.
7. Esposito ML, Morine KJ, Annamalai SK, et al. Increased plasma-free hemoglobin levels identify hemolysis in patients with cardiogenic shock and a trans valvular micro-axial flow pump. Artif Organs 2019.43: 125–131.
8. Tanawuttiwat T, Chaparro SV. An unexpected cause of massive hemolysis in percutaneous left ventricular assist device. Cardiovasc Revasc Med 2013.14: 66–67.
9. Gaudard P, Mourad M, Eliet J, et al. Management and outcome of patients supported with Impella 5.0 for refractory cardiogenic shock. Crit Care 2015.19: 363.
10. Scanes CG. Sturkie’s Avian Physiology, 2015.6th ed. Elsevier.
11. Strunk A, Wilson GH. Avian cardiology. Vet Clin North Am Exot Anim Pract 2003.6: 1, v–28, v.
12. Fraser KH, Zhang T, Taskin ME, Griffith BP, Wu ZJ. A quantitative comparison of mechanical blood damage parameters in rotary ventricular assist devices: Shear stress, exposure time and hemolysis index. J Biomech Eng 2012.134: 081002.
13. Molteni A, Masri ZP, Low KW, Yousef HN, Sienz J, Fraser KH. Experimental measurement and numerical modelling of dye washout for investigation of blood residence time in ventricular assist devices. Int J Artif Organs 2018.41: 201–212.
14. Taskin ME, Fraser KH, Zhang T, Wu C, Griffith BP, Wu ZJ. Evaluation of Eulerian and Lagrangian models for hemolysis estimation. ASAIO J 2012.58: 363–372.
15. Ballyk PD, Steinman DA, Ethier CR. Simulation of non-Newtonian blood flow in an end-to-side anastomosis. Biorheology 1994.31: 565–586.
16. Gnecchi-Ruscone T, Rigo F. Normal range for pulmonary artery systolic pressure. JAMA 2008.299: 2022; author reply 2022–2022; author reply 2023.
17. Moazami N, Fukamachi K, Kobayashi M, et al. Axial and centrifugal continuous-flow rotary pumps: A translation from pump mechanics to clinical practice. J Heart Lung Transplant 2013.32: 1–11.
18. Song BW, Wang YJ, Tian WL. Open water performance comparison between hub-type and hubless rim driven thrusters based on CFD method. Ocean Eng 2015.103: 55–63.
19. Song X, Wood HG, Allaire PE, Antaki JF, Olsen DB. Inlet and outlet devices for rotary blood pumps. Artif Organs 2004.28: 911–915.
20. Kameneva MV, Burgreen GW, Kono K, Repko B, Antaki JF, Umezu M. Effects of turbulent stresses upon mechanical hemolysis: experimental and computational analysis. ASAIO J 2004.50: 418–423.
21. Ozturk M, O’Rear EA, Papavassiliou DV. Hemolysis related to turbulent eddy size distributions using comparisons of experiments to computations. Artif Organs 2015.39: E227–E239.
22. Thamsen B, Blümel B, Schaller J, et al. Numerical analysis of blood damage potential of the HeartMate II and HeartWare HVAD rotary blood pumps. Artif Organs 2015.39: 651–659.