Conducting finite-element analysis on new and redesigned medical components can save precious time and money during FDA approval.
Stephen J. Mraz
Autodesk Algor, a subsidiary of Autodesk, www.autodesk.com/algorsimulation
Saba Metallurgical and Plant Engineering Services LLC, www.smpes.com
A company that developed a new line of intervertebral body-fusion devices (IBFDs) almost shelved the entire project when confronted with FDA-testing requirements. The FDA mandates that all companies must test physical prototypes to ASTM standards before a device earns approval. The new line consisted of three separate but similar designs for distinctly different areas of the lumbar region of the spine. It also had six to eight size variations to better match patient anatomy. With the tab for physically testing one device coming in at around $50,000, testing the entire line would cost $1.2 million and make the project uneconomical.
To get around this financial hurdle, the company hired Saba Metallurgical and Plant Engineering Services LLC (SMPES), a consultancy based in Baton Rouge, La., to carry out FEA on its designs. The intent was to show that the company’s weakest IBFD, at least in terms of FDA testing criteria, would still pass FDA testing. Therefore, all the other devices, which would be stronger, would also pass FDA testing and needn’t go through expensive physical testing.
Lumbar (lower-back) implants mainly experience compression loading both statically and in fatigue, so the FDA demands they pass static axial compression and axial fatigue tests which measure the device’s strength. This is in contrast to cervical (neck) implants which see compression and twisting, so the test regime for cervical devices includes torsion. But note that none of the tests are designed to reveal how effective the medical device will be in its task, just that it will be strong enough to withstand the loads and stresses normal human activity will place on it.
Brent Saba, chief engineer at SMPES, relied on various FEA procedures he has developed in his work on orthopedic implants to model tests. These procedures ensure results closely match those of actual physical testing. For example, he has painstakingly developed procedures that cover mesh-density convergence tolerance and degrees of freedom. One of his most valuable resources, however, is his database of material properties.
IBFDs must also be tested for subsidence and expulsion. Subsidence (ASTM F2267-04) is a measure of how deep a harder material will press into softer one. Too much subsidence in an IBFD and the space between healthy discs decreases, which curves the spine unnaturally. This can make support hardware such as plates and rods become unstable. Physical testing involves pressing an IBFD into a standard polyurethane foam block and measuring the depth of the impression.
|Why Algor FEA?|
Brant Saba, founder of Saba Metallurgical and Plant Engineering Services, relies on FEA for his work on spinal implants as well as projects involving pressure vessels, tanks, and piping. He has been a long-time user of Algor’s FEA software for several reasons. (Algor is now a subsidiary of Autodesk.) “The company is happy to work with small engineering firms like mine,” says Saba. “Other companies didn’t even return my calls requesting quotes.” Saba also appreciates the user-friendly aspect of the programs and, most of all, its accuracy.
Expulsion measures how well the device remains in place despite a force trying to dislodge it. There is no standard ASTM test for expulsion. So developers usually place an IBFD between two blocks, apply a slight clamp load on the blocks, then use controlled displacement to push on the IBFD until it moves a specified distance.
FDA sometimes requires IBFDs be shear tested. And if the IBFD has more than one component, the FDA will also require wear testing and possibly a study of screw pullout.
The two greatest problems in setting up accurate FE analysis of these implants was to come up with a proper mesh and finding material data that can be trusted.
“Static axial-compression tests called for relatively large equiaxed elements that can handle high levels of distortion,” says Saba. “Small and acute elements can get too distorted and never let the FEA program finish solving. The difficulty with IBFDs is with tight curves and closely spaced openings. So much of my time is spent on getting the proper mesh.”
For static analysis, Saba employs strain (displacement) controlled loading to get a load versus displacement graph. He sums the nodal reaction forces at the loading surface and plots them. From this second graph, he derives the 2% offset yield which is directly compared to the results from physical testing.
So far, his results with titanium-based devices have been successful when it comes to getting an ultimate load strength. But analyzing devices made of PEEK is a bit more problematic. “FEA generates a much-lower ultimate load strength than physical tests because FEA software cannot handle all the compression,” says Saba. “And the FEA programs usually stop solving if there is too much buckling. In the lab, the device can be compressed much more even after a buckling event.”
Another problem is that short devices, those less than 5-mm tall, can act like gaskets and actually increase in strength as displacement is reduced. “This calls for exceptional compression/distortion modeling and is not possible at this time.”
Saba also doesn’t rely on plastic-collapse analysis for measuring fatigue. Lab results show that models must have a high level of refinement to get accurate peak-stress levels, according to Saba. And this calls for a high-density mesh with anywhere from 100,000 to 1 million elements. “So plastic-collapse analysis is too memory intensive with that many elements. Besides, the stresses found using it are nearly always less than the material yield strength, so with stresses in the elastic-stress range, the stress results remain quite accurate.”
The other challenge is getting good material data. “Most manufacturers provide few material proprieties, particularly for yield strength,” notes SABA. “And PEEK can vary widely from batch to batch when it comes to yield strength.”
In the end, FEA let Saba compare IBFDs of differing shapes and sizes, letting him determine the weakest shape and size combination. After this weakest combination passed both FEA and physical testing, the FDA did not need the same done on all shapes and sizes. And should a device fail FEA testing, engineers can refine the design and retest it using FEA. “Catching potential flaws during the design state instead of after machining prototypes that would later fail in lab testing provided tremendous savings in cost and time,” says Saba