Skip to:

CFD for Blood Transfusions on the Battlefield and Inhalation of Toxic Agents in the Lung

Eric E. G. Shaqfeh, Stanford University
Gianlucca Iaccarino, Stanford University
Armando Rodriguez, Army Institute of Surgical Research
Andrew Cap, Army Institute of Surgical Research
Lorne Blackbourne, Army Institute of Surgical Research
Jana Kesavan, Edgewood Chemical Biological Center
Steven Hill, Army Research Laboratory
Paul Dabisch, National Biodefense and Countermeasures Center

This project has two components.  One component is to study the adhesion of blood platelets to an injured vessel site. This is a critical initial stage for the formation of a platelet plug to stop bleeding.  The second component is to study the deposition of aerosol particles in the lungs to help study the effects of airborne pollutants, and infective and toxic agents.

Blood flow for trauma-related applications: Red blood cells play a critical role in influencing human bleeding time. Trauma remains the leading cause of mortality for soldiers in combats, and 25% to 35% trauma victims experience an initial bleeding diathesis upon presentation to a medical facility. In a bleeding event, blood vessel walls are damaged, which trigger platelets to travel to the site of injury and start the clotting process. Surprisingly, previous studies show that bleeding time is influenced more by the fraction of red blood cells (hematocrit) than the fraction of platelets.

In this project, we build a computer model to understand the physical chemistry of blood flow. Our findings offer insights into the mechanics of trauma injuries and blood disorders such as sickle cell anemia and malaria. Our computer model offers a quantitative explanation for how red blood cells can influence the near-wall concentration of platelets, and thus affect the traveling time of platelets in the event of bleeding. 

Our study is composed of building a high fidelity computer simulation of blood flow and deriving a simplified model for fast computations. This approach does not require the use of large volume of blood in conventional lab studies and avoids the noises and degradation in experimental samples. We also work closely with Army researchers who conduct experiments, as shown in Figure 3, to measure quantities such as the platelet adhesion rate. We can thus verify our model by comparing with those experimental data.

In the first part of our study, we utilize army’s supercomputing resources to develop a set of large-scale computer simulations.  We model red blood cells as deformable particles and platelets as rigid disks. Each simulation contains hundreds of particles and matches the geometry and flow conditions in actual arterioles and venules. The shape change of red blood cells agrees with experimental observations.

As shown in Figure 4, we report the increase in cell-free layer thickness when the hematocrit is reduced. As a result, the near-wall accumulation of platelets is less pronounced and leads to prolonged bleeding time.

Our simplified model serves as a fast alternative to large-scale simulations. In this scaled-down version, we look at the average behaviors of red blood cells and platelets instead of predicting the behavior of each individual particle. We are the first group to derive such a model and extract key variables relevant to bleeding. In the treatment for trauma injuries, saline is usually injected, which reduces the hematocrit. With our model, we can estimate the cell-free layer thickness at a given hematocrit. Thus, this leading-edge technique enables us to do individual-based prediction of bleeding time on a personal laptop. 

In addition to healthy red blood cells, we also study how red blood cells with varying shapes, sizes, and deformability -- including the crescent-shaped cells that are the hallmark of sickle cell anemia and the rigidified cells invaded by malaria-- will influence blood flow. These variations influence the cell-free layer thickness and have a broader impact on hemostasis and thrombosis.

Figure 4: snapshots of the blood simulations at 15% (top) and 20% (bottom) hematocrits.

We are currently working on a new simulation method that is faster and more flexible. This powerful tool will open up exciting opportunities for future research. For example, we can simulate more complex geometries such as the branched blood vascular network. We can also include various micron- and nano- sized particles in the simulations, which Army researchers are interested in for drug delivery applications.

The collaboration between Dr. Shaqfeh's group at Stanford University and the Coagulation and Blood Research Program at the US Army Institute of Surgical Research has been very productive. The simulations of hematocrit effects on platelet adhesion are corroborated by experimental findings and suggest strategies for improving transfusion strategies in the management of combat casualties. We look forward to continuing this important collaboration.  –LTC Andrew Cap,  US Army Institute of Surgical Research

Inhalation of toxic agents in the lung:

Soldiers on and off the battlefield are exposed to dangerous substances in the air they breathe. Airborne debris and smoke from explosions may contain radioactive or toxic compounds. Soldiers may also be at risk from biological threats in the air such as anthrax or viruses like Ebola. The lung and its extended network of airways provide a portal of entry for these airborne contaminants.  Lung particle deposition is defined as the fraction of particles that do not deposit during inhalation and exhalation. The ability to predict this deposition is therefore critically important for understanding the fate of inhaled aerosols and its effect on pathogenesis in soldiers. The current project aims to develop HPC simulation capabilities to study the flow in the lungs and the deposition of minute particles. Detailed information about particle deposition in the lungs will enable more effective measures to protect human health.


Figure 2. LES simulations in a human airway (left) from mouth to generation 7 of the lung show remarkable accuracy when compared to MRV experiments

To study inhalation, we developed a computer code that is capable of giving detailed airflow and particle deposition information in the lungs. The “Virtual Inhaler” code was developed in-house at Stanford with the objective to provide all of the necessary capabilities required by this complex application, including turbulence modeling and scalability. It is capable of predicting air dynamics in any lung. Given any person’s CT or MRI scan, it can compute that person’s specific breathing dynamics from the mouth and nose to generation 7 of the lungs. The code is also optimized for parallel computation, allowing it to run quickly and efficiently on massive supercomputers. Scalability of the computations is of great importance because of the geometrical complexity of the realistic airways, which require extremely large computational meshes. The separation-of-scales from the large trachea to the tiny alveoli requires us to develop low-order approximations to accurately model deposition in millions of small passages.

We have developed Large Eddy Simulation (LES) cases for the flow field in three human airways.  The first is an idealized model constructed to represent a statistically average human lung and provides generalizable results. We have also produced results in a healthy middle-aged male lung and the lung of a person diagnosed with sleep apnea. Additionally, in collaboration with Army researchers who perform animal inhalation experiments, we have also computed flow and deposition results in five Rhesus Macaque monkey lungs to better understand how experiments in monkeys compare to human lungs.

We have validated flow results with experimental measurements of the 3D velocity field obtained with Magnetic Resonance Velocimetry (MRV) scans carried out in the lab of Prof. John Eaton (Dept. of Mechanical Engineering, Stanford). The realistic geometry was used to design and manufacture a 3D printed human lung. Time-averaged velocity field data was obtained by MRV experiment and simulation. A detailed comparison showed a high level of agreement which is unprecedented in inhalation simulations and gives us a high degree of confidence in the ability of the “Virtual Inhaler” to model airflow.

Finally, in collaboration with Army researchers at the Edgewood Chemical Biological Center (ECBC) we have computed LES results for 5 Rhesus Macaque monkeys. Rhesus monkeys are commonly used as a stand-in for humans in determining the toxic effects of inhaling various toxic particles (such as anthrax spores or the Ebola virus). It was previously unknown whether turbulence would develop in the monkey lung due to the smaller size and lower inhalation flow rates. Our modeling efforts have convincingly proven that turbulence does form in the larynx of the Rhesus monkey. Dr. Jana Kesavan at ECBC performed particle deposition experiments by forcing particle-laden air through 3D printed models of monkey airways. Our simulations showed near-quantitative agreement with the Army experiments, and provided detailed flow field and deposition patterning results that are impossible to obtain by experiment. This result is currently being compiled into joint publications with researchers at ECBC and will be submitted in the coming months.

The Aerosol Sciences Team at Edgewood Chemical Biological Center has been collaborating with Dr. Eric Shaqfeh's group at Stanford University to determine the amount and location of particle deposition in animal and human respiratory systems.  The ECBC group has been experimentally determining the particle deposition in anatomically correct 3D printed models while Dr. Shaqfeh's group has been mathematically modelling the particle deposition.  Additional testing has been conducted using simple structures to validate the mathematical modeling.  Validated mathematical modelling will allow predictions of particle deposition in various animal and human respiratory systems and in other tubes and ducts.  -Dr. Jana Kesavan, ECBC