Bookmark and Share

Integrated Artificial Pump Lungs for Respiratory Failure

Computational Fluid Dynamics

The Medtronic Affinity NT Oxygenator
The Medtronic Affinity NT Oxygenator

Computational Fluid Dynamics measures pressure drop and oxygenation in blood membrane oxygenators. The blood oxygenation process in hollow fiber membrane-based blood oxygenators includes: (1) transport of the oxygen from the lumen of the hollow fibers across the fiber wall into the blood plasma and red blood cells by both diffusion and convection; (2) oxygen binding with blood elements, i.e. hemoglobin, and (3) oxygen releasing back to plasma from hemoglobin during the dynamic oxygen transfer process. To model the above process, a mass transport equation for the oxygen transfer to plasma for the blood oxygenators was developed from the standard convection-diffusion mass transfer equation. The standard equation was modified to accommodate the specific transport processes associated with hollow fiber membrane-based blood-gas oxygenators, including oxygen consumption from the hollow fiber membranes, oxygen binding with blood and oxygen reacting with hemoglobin.

The governing equation was programmed and then compiled into Fluent 6.1 solver using user-defined subroutines (UDF) and user-defined scalars (UDS). It was then coupled with continuity and momentum equations to solve the velocity and pressure fields, gas transfer, oxygen partial pressure and O2 saturation.

The Medtronic Affinity NT oxygenator with plasma resistant fiber (PRF) was used as our test device for computational simulation (Medtronic Inc., Anaheim, CA, U.S.A.). The Medtronic Affinity NT oxygenator utilizes a graduated bundle density design that maximizes gas transfer, while blood shear, priming volume and blood-side pressure drop are minimized. Twelve access ports were put on the outer casing wall around the fiber bundle of the Medtronic Affinity NT oxygenator to measure the fluid pressure, oxygen partial pressure and saturation. Three sets of four access ports, each from top to the bottom, were placed along front, back, and one side of the oxygenator relative to the outlet port.

In vitro experiments were performed in a mock flow loop using both saline (0.9% sodium chloride) and bovine blood as the testing fluids.

fluid-pressure distribution (units in mmHg) in the oxygenator
The fluid-pressure distribution (units in
mmHg) in the oxygenator at the flow
rate of 5 liters/min.


The fluid-pressure distribution (units in mmHg) in the oxygenator at the flow rate of 5 liters/min for bovine blood is presented in the figure below. The pressure contour correlates well with the velocity field.

The blood flows upwards through the core of the mandrel and is then redirected back downward and distributed in the gap between the outer surface and the inner surface of the hollow-fiber bundle. The flow radiates outwards through the fiber bundle, circulates in the space between the outer periphery of the fiber bundle and the casing wall towards the outlet port, and then exits through a single port at the lower edge of the fiber bundle. The pressure in the inlet is higher than that in the outlet port, but the pressures in these regions are relatively uniform. This was verified by the experimental measurements from the pressure ports drilled on the oxygenator axially.

The numerically predicted oxygen partial pressure and saturation correlated well with the experimental measurements, as presented in the figure below. The difference of oxygen (O2) distribution profiles between using saline and bovine blood as testing fluids is significant.


These experiments indicate that water can’t be employed as a substitute for blood to characterize the oxygen transfer in biological devices. This is due to the different gas-exchange mechanism. However, glycerol water can be used to simulate blood in hydrodynamic analysis when adjusted to the same viscosity of blood.

Numerical Prediction of Shear Stress-Induced Hemolysis

The model adopted here is based on the same idea proposed by Garon and Farinas, which can be developed and written as a hyperbolic equation. The three-dimensional shear stress field was calculated from the velocity field obtained from the numerical simulations of the blood flow in the oxygenators. The stress tensor is reduced to a scalar as originally used by Bludszuweit.


The Medtronic Affinity NT and COBE Optima XP were used to test the mathematical model adopted in this study. Section views of the velocity field and scalar stresses are depicted in the figures below. The simulated results compared well with the experimental results from Kawahito S. et al.


The velocity profile and localized scalar stresses are used to indicate high-risk regions for blood trauma. The section view was taken from the symmetry plane of the oxygenator along the mid-plane of the inlet. It shows the detailed flow paths and the magnitude of velocity, which reduces dramatically in Fiber buddle. The streamlines illustrate the radial flow paths in both Affinity NT (a) and COBE Optima XP (b) oxygenators.

The current model eliminates the need to track the particle paths in Lagrangian coordinates, which could cause erroneous predications and computational difficulties when the velocity nears zero, or in complex geometries. Another advantage is that the equation used in this paper can be programmed with continuity and momentum equations in any coordinate system. It is solved in the whole domain, and can pinpoint the location of high lysis inside the computational domain of the medical devices.

Development of the Pump Lungs

We are currently developing a series of integrated pump-lung devices, including bedside and wearable ambulatory devices, for patients with acute and chronic lung disease. Shown in the picture below is a newly developed fully integrated bedside pump-oxygenator.

The device combines a magnetically levitated centrifugal pump and a hollow fiber membrane bundle to form one single compact system capable of both pumping and oxygenation. The pumping function of the device was based on the magnetically levitated bearingless impeller/motor technology. The only moving component within the pump is the impeller, which rotates in a contact-free manner. The drive, magnetic bearing and pump rotor functions are incorporated into a single unit, eliminating all valves, seals, mechanical bearings or other moving parts. This minimizes the risks of hemolysis, thrombus formation and mechanical failure associated with traditional mechanical bearings and valves.


Initial animal experiments with the bedside device have demonstrated very promising results. The device exhibited excellent gas transfer performance, biocompatibility and durability over 12 days. There was no plasma leakage into the fibers or thrombosis in the flow path and fiber bundle.

Evaluation and Development of Non-Thrombogenic Fiber Coatings

It will be critical to minimize platelet deposition onto the fibers within the pump-lung device to prevent membrane fouling and device failure as well as to minimize patient morbidity. Heparin coating has been proven repeatedly to minimize the thrombogenecity of cardiovascular devices. We have access to a proprietary heparin-albumin coating that has been used commercially. The biocompatible coating is a multi-layer coating comprised of alternating layers of human albumin (Bayer Human Albumin 25%, Bayer Corporation, Berkeley, Calif.) and sodium heparin (Celsus Laboratories, Inc., Cincinnati, Ohio). A crosslink (1-Ethyl 3-(3-dimethylaminopropyl)-carbodimide) is applied to stabilize the heparin molecules onto the albumin-coated surface. The coating process is performed after device fabrication but before sterilization.

In-Vivo Evaluation of Functionality, Biocompatibility, Hemocompatibility and Durability in Animals

We have performed five in-vivo animal experiments with the implantable pump-lung device based on the spinning disk of fiber membranes. The device was surgically connected between the atrium and pulmonary artery (PA) of Jersey calves. The implantable pump-lung device utilized both uncoated and silicone-coated HFM and had a surface area of 0.5 m2. The mean blood flow rate was approximately 4 Lpm.

The sweep gas-flow rate was maintained at 1:1. The time-averaged value of the normalized oxygen transfer rate in a one-day study remained above 350 ml O2/min/m2. The oxygen gas transfer from the five-day study varied over the duration of the study. The mean oxygen transfer rate was about 110 ml/min over five days. The oxygen transfer rate did not deteriorate with time. Plasma-free hemoglobin was increased from a pre-operative level of 5 mg/dL to 28 mg/dL post-operative and remained elevated over the period of 5 days. The platelet and white blood cell counts remained in the normal physiological ranges. Necropsy results exhibited that there were no infarcts in the lung for any of the five animal studies.