How to speed up the 60601 drop test with finite element analysis

Figure 1: From 3D CAD model (left) to FEA mesh model [Image courtesy of Flex]

Finite element analysis can help design a robust mechanical architecture to pass one of the toughest tests in medical device design.

Giorgio Sardo, Flex

Time to market is a key factor in the success of a new medical product. Designers now use modeling and simulation to pass one of the toughest tests: 60601-1, an international standard applicable to all medical electrical equipment and systems. By avoiding costly iterations of trial and error, they can quickly define the parameters of a robust mechanical architecture. Drop simulation can predict the test result and also support failure analysis investigation.

If you’ve ever dropped your phone, your immediate concern is whether it’s still working. Medical device users do not feel the same anxiety if they drop their device because medical electrical equipment must meet a set of strength requirements that ensure the device will work, even after an accidental drop.

Devices ranging from small glucometers to huge CT scanners must comply with IEC 60601-1, which ensures that equipment is designed to maintain basic safety and essential performance.

In the regulatory world, this can be a complex issue requiring experts to agree on precise definitions. For example, one could say that basic safety is ensured if the device does not present an unacceptable risk when it is used normally or when a single element fails; essential performance can be considered as the function which, if lost, would produce an unacceptable risk.

From a mechanical point of view, a device must be strong enough to prevent any loss of basic safety or essential performance caused by rough handling or excessive force resulting from impact or dropping. Body-worn, hand-held and portable devices are drop tested based on the average of three free drops from a height of at least 1m onto a 50mm wooden plate resting on a base rigid concrete or similar.

Two ways to pass the drop test

This is a difficult test with two paths to success:

  • Trial and error: This can be a time-consuming and expensive option that requires producing test samples, evaluating each test performance, analyzing failures, and redesigning another iteration for testing up to that the requirements are met.
  • The alternative is Finite Element Analysis (FEA). A device is discretized into several small pieces and then the deformation, stress and acceleration of the product components are checked after the simulated ground impact.

Before we continue, let’s clarify what we mean by parts used to mesh the model. These are elements into which the 3D model of the device can be divided (see Figure 1); they can be one-dimensional, two-dimensional or three-dimensional, include tetrahedrons and hexahedrons and be connected to each other by nodes located respectively on each of the four or eight vertices. This mesh can be made automatically by the software, but also refined and modified by the user according to the FEA objective.

There is more than one software capable of performing this analysis. We will consider ANSYS Explicit STR, which exploits the advantages of an explicit solver and facilitates the management of nonlinearities which are contact points between different parts.

ANSYS also provides an implicit module which is normally used for static analysis or when the duration of the simulated event is relatively long (seconds or more). In this case quadratic elements can be used as they have intermediate nodes and vertices and can provide a more accurate solution.

This module is used to calculate stresses and strains from the interaction of two (or very few) parts, because the management of nonlinear contacts with and without friction is relatively complex using an implicit iterative numerical scheme.

The explicit module, on the other hand, is generally used for very short events (milliseconds) and can handle models with multiple parties and associated contacts. This means that the entire device can be modeled and verified. The explicit approach follows a chronological path, therefore performs a series of cycles where the solution of time step “I” depends and comes from the solution of time step “i-1”, so no iteration and no convergence check are necessary.

It is inherently unstable and very short time steps are needed to make it stable. That’s why it’s so good for very short phenomena and when you need to look at what’s happening in very short time steps, like fractions of microseconds.

Regarding the mesh, ANSYS Explicit STR uses linear and not quadratic elements. Consequently, a refinement of the mesh is often necessary in the area of ​​interest. Tetrahedra are generally preferred over hexahedra for complex geometries such as in plastic parts used in medical devices, and the mesh diagnostic tool provided by the software allows checking the quality of the mesh before launching the simulation.

A main step before and often after meshing is the simplification of the model. Very complex features such as screw threads or corner roundings will be removed whenever it does not affect the simulation results. It is important to remember that a finite element model is always a simplified version of reality and that achieving the best balance between simplification and reliable results requires experience.

Now let’s look at the fall simulation. Only a few milliseconds can be assessed — depending on the computing power available — so the mesh model is placed just a few hundredths of a millimeter from the simulated ground floor. The gravitational acceleration velocity produced in 1 m of running is applied to the falling body as the initial condition. We then define the outputs to be checked, for example strain and stress of components or features.

During the simulation, the time step is automatically adapted according to the deformation of the mesh elements. To avoid a runtime crash, you can remove the bricks that deform too much, while checking the quantity and location of these eroded elements. This is important because we don’t want to affect the physics of the problem and cheat. The alternative is to fix the problem in the meshing process – go back to pre-processing.

Von Mises stress (Mpa)

Figure 2: Von Mises stress (Mpa) [Image courtesy of Flex]

Data post-processing

The last step is the post-processing of the data. Information can be extracted and traced to understand if and where criticalities are occurring. For ductile materials, the key indicator is strain, while for more brittle materials, stress – specifically yield strength – is the point of interest (see Figure 2). In some other cases, acceleration may be relevant.

At other times, acceleration may be relevant. In this case, the drop simulation involved understanding whether a crystal soldered to the device PCB (Figure 3) was damaged due to the drop or because it was a weak sample.

Broken crystal resonator and position of the crystal on the circuit board

Figure 3: Broken crystal resonator (top) and position of the crystal on the circuit board [Image courtesy of Flex]

From the CAD model, all parts of the device were simplified and then meshed. Material properties were assigned to each part and connections defined, then the electronic component was modeled as a plastic part and glued to the PCB.

The simulation lasted 2ms and it was enough to see the first impact and the beginning of the rebound. An acceleration probe was placed on the component considered to be a rigid body and the acceleration plotted (Figure 4).

This result was compared to the specification provided by the data sheet of the electronic component which specifies that the maximum authorized acceleration is 5000G for 3ms ½ sine wave. Looking at the data above, one could conclude that the 5000 G for 3ms ½ sinusoidal was not reached. Additionally, looking at the higher frequency results (10 kHz), an 11,000 G spike can be identified as lasting only a very short time (~0.05 ms) well below the limit identified by the vendor. .

Figure 4: Result of average acceleration on a crystal rigid body (G vs ms), vertical direction.
The resulting upper graph was very noisy and filtered using an ANSYS function with a 2kHz filter producing the lower graph. [Image courtesy of Flex]

We could conclude that the component failure was due to weakness in that specific sample and, once verified, avoid a time-consuming and costly redesign. We can clearly see that even after the simulation, we can find cost and time savings and increase the efficiency of the design process.

This example clarified how finite element analysis helps us understand the results of a drop test. It demonstrates that several aspects can be taken into account and how the creation and testing of several samples can be avoided.

The modeling phase is the key factor for reliable results. This requires experience and a vision of the final target, but takes much less time than building and testing multiple samples. This approach provides direct savings on the project schedule and budget, and a competitive advantage for the manufacturer who can get their product to market faster.

Giorgio Sardo is Principal Design System Engineer at Flex [Photo courtesy of Flex]

Giorgio Sardo, Senior Design System Engineer at Flex, has worked at the Flex Milan Design Center for over 15 years. Sardo has extensive experience in the design, prototyping and industrialization of various types of medical devices, such as auto-injectors, injection pens, glucometers, portable drug delivery devices and dental tomograph.

The opinions expressed in this article are those of the author alone and do not necessarily reflect those of or its employees.

Source link

Comments are closed.