you are here:   / Resources / The Use of LS-DYNA Models to Predict Containment of Disk Burst Fragments

The Use of LS-DYNA Models to Predict Containment of Disk Burst Fragments

Turbomachinery manufacturers commonly test the centrifugal strength of their rotors in a vertical axis spin test, often called a disk burst test. The design of the containment shell that encloses the disk burst event is critical to ensure the safety of the area surrounding the test. One method used to design the containment shell for a turbine disk burst test is based on the assumption that the kinetic energy lost by the disk fragments during impact is converted into kinetic energy in the containment shell and energy loss to plastic strain and shear failure in the shell.

Containment shells are sized such that the energy required to fail the shell material exceeds the kinetic energy loss during impact. This method is described in an ASME paper entitled "The containment of Disk Burst Fragments by Cylindrical Shells," by A.C Hagg and G.O. Sankey (ASME publication 73-WA-Pwr-2).

These methods are approximate because they assume fully inelastic impacts and do not account for losses due to friction or heat, nor do they account for stress concentrations in the impact zone or complex disk geometries. A predictive analytical tool for predicting disk containment would provide enormous benefit and remove some of the conservatism and extra cost inherent in using traditional guidelines.

CAE Associates used the ANSYS/LS-DYNA software to develop an analysis method that our client could use in order to make more accurate assessments of containment failure limits, thereby optimizing the design of the containment shell for particular disk geometries and spin speeds.

The ANSYS/LS-DYNA explicit finite element transient dynamics analysis employed a piecewise linear plasticity material law with strain rate dependence to represent the nonlinear material behavior of the containment shell.  In addition, segment-based eroding contact was defined between the disk fragments and the containment shell in order to be able to simulate penetration of the disk through the shell.

Simulation sensitivity parameters included failure strain, mesh density, friction, hourglass control, time step size and the inclusion of double precision.

Initial studies showed that results were very dependent on mesh density, which is common in damage-based models. However, it is important to try to minimize this influence if the analysis is to be used in a predictive fashion. To get around this difficulty, CAE Associates implemented a non-local failure method. Once this method was employed, the analysis results were less sensitive to mesh density and provided very consistent failure predictions across various test geometries, loadings and material properties.

The methods developed in ANSYS/LS-DYNA for predicting containment were compared to the burst test results presented in the aforementioned paper and the final calibrated models showed good correlation with test data relative to the prediction of containment, shell perforation, and overall deformation.

The resulting modeling methods and calibrated parameters provide a tool that our client can use to predict containment of disk burst fragments which relies on fewer approximations than earlier energy-based methods. This provides the ability to better tailor the design of the containment shell for specific disk geometries, thereby reducing costs due to excess material and weight.

The details of this work can be found in a paper prepared by CAE Associates, entitled  “The Use of LS-DYNA® Models to Predict Containment of Disk Burst Fragments” (Click here to view the paper.) and presented at the 10th Annual LS-DYNA Users’ Conference.