Design and operational challenges of wearable drug delivery systems.
Figure 1 summarizes the various engineering problems that are going to be addressed. At a high level, we need to make sure that the drug delivery to the patient is accurate and reliable. To do that, we have to worry about six themes:
1. Drug delivery: Catheter design, drug flow, kinking potential (establish and maintain the flow path), drug infusion.
2. Pump durability: Drop testing.
3. Antenna design and placement: Wireless communications and interference.
4. Power management: Low-power system.
5. Embedded software development: For the controller and the display.
6. Virtual systems prototyping: While we can look at each of the above in isolation, it is the multi-physics, multi-scale, systems level understanding of this device that will ensure reliability and performance.
The system's "V" development model
Figure 2 shows the typical "V" process for developing a medical device (or any product for that matter). The process begins with the marketing requirements on the upper left. We then start designing the system and its components. At the bottom of the V is where bench testing and simulation is being performed on each of the individual components (e.g., catheter extrusion, kinking prediction, insulin flow and diffusion). Then we start accounting for interaction of the various components as we go up the right hand side. Component integration (one level up from the bottom) is more or less enabled by the multi-physics modeling capabilities, allowing users to easily pass results between computational fluid dynamics (CFD), structural analysis (FEA) and electromagnetics (emag) analyses. But then, as we go further up the chart, we need to start including the device in its intended environment, and that is where we see the requirements for systems simulation (for medical devices and all other products that are out there).
Designing the system
The real challenge is to develop a systemlevel understanding of the system (insulin pump in this case) behavior, which helps engineers understand how components and controls behave when assembled into a fully integrated system. In this case, the entire system behavior is modeled using the ANSYS Simplorer tool. ANSYS Simplorer allows multi-domain models composed of electrical, hydraulic, mechanical and other physics to be assembled into a unified system that can be used to verify that the optimum components also make for an optimum system. The multi-domain models might be reduced-order models (ROMs), like what is represented in figure 3. Figure 3 shows an illustration of the system model of the insulin pump, which is comprised of embedded software for the display and motor controller, power electronics to drive the motor, and the hydraulics for modeling insulin delivery. As an example, let us focus on the hydraulics system.
To be able to size the pump correctly, one needs to know the pressure loss that it needs to overcome to deliver the required volume of insulin. Knowing the tubing length, diameter, and insulin viscosity and density, the pressure loss can indeed be predicted through back of the envelope computation. However, one common scenario that complicates this analysis is the kinking phenomenon (figure 4). If kinking is to occur, it will dramatically reduce the area available for the insulin to flow through, and hence pose a substantially higher flow resistance to the pump. Additional action will be needed to ensure that enough insulin is still delivered, or the user needs to be notified to correct for the kinking.
ANSYS Mechanical is used to simulate the kinking process (figure 4), taking the catheter dimensions and material into account. The geometry of the kinked tube is exported, and the insulin flow through the deformed geometry is simulated. A design of experiments (DOE) is run to obtain flow characteristics (pressure drop versus volume flow rate) at various kink angles (figures 5 and 6).
The ROM for the pressure drop versus flow rate is then used to develop the hydraulics system model shown in figure 3.
Embedded software system
For the embedded software piece, designers want to understand the behavior of the embedded software in the intended use environment. For this insulin pump, the displays and control systems developed in ANSYS Scade can be integrated into the Simplorer model (hydraulics system) so that users are able to test the code in a "software in the loop" type of configuration. This can help engineers determine if the logic, controllers and displays will work, even before any hardware is built.
Typically, the geometry is simple (single lumen cylindrical catheter). However, the small size and the strict dimensional requirements do pose a manufacturing challenge. Also, this challenge is magnified by the fact that typical materials used (e.g., HDPE) exhibit a non-linear behavior (viscoelasticity) that is responsible for the die swell phenomenon (this is when the material expands as it leaves the die). From a manufacturing perspective, the question is what should the size of the die be that will result in the required final shape and dimension. In addition, typically draw down is applied during the extrusion process as another means of controlling the dimension, as well as making the die cavity easier to manufacture (due to draw down, the size of the die can actually be much bigger than the final size). In this case study, ANSYS Polyflow was used to predict what is the die opening size needed to reach the final catheter diameter. This simulation is called "inverse extrusion," where the simulation outcome is the die opening size for a prescribed tube size (figure 7).
The design, operation and manufacturing of a typical insulin pump has been considered. Different system level models were discussed with emphasis on the hydraulics system for the insulin delivery, considering the possible kinking of the catheter. A reduced order model (ROM) was created for the pressure drop versus flow rate that was then placed into a zero order hydraulic model. In addition, an embedded software system is also developed to drive the hydraulics system to allow for real time model simulation.
This article is based on a paper originally presented at SPE's Thermoplastic Elastomers 12th Topical Conference, September 2016.
by Hossam Metwally, Ashutosh Srivastava, Marc Horner, Aleksandra Egelja-Maruszewski, Fadi Sawaged and Mark Solvenson, ANSYSInc.
Caption: Figure 1--engineering problems involved in a typical insulin pump design and operation
Caption: Figure 2--"V" systems development model
Caption: Figure 3--insulin pump system illustration; embedded software (display of motor and controller), power electronics (electric motor drive) and hydraulics (insulin delivery)
Caption: Figure 4--kinking phenomenon, sudden reduction of flow cross-sectional area
Caption: Figure 5--pressure drop through the catheter at different kink angles
Caption: Figure 6--reduced order model (ROM) for pressure drop vs. flow rate at different kink angles
Caption: Figure 7--inverse extrusion simulation
|Printer friendly Cite/link Email Feedback|
|Author:||Metwally, Hossam; Srivastava, Ashutosh; Horner, Marc; Egelja-Maruszewski, Aleksandra; Sawaged, Fadi;|
|Date:||Jun 1, 2017|
|Previous Article:||Selection and application of antimicrobial additives in thermoplastic compounds.|
|Next Article:||Bonding solutions for styrenic block copolymer TPE medical tubing.|