NASA Ares I
HyperSizer was selected for use on the NASA Ares HLV project. New capabilities were programmed into HyperSizer specifically in support of the Ares upper stage project.
HyperSizer Orthogrid Test Panel Pretest Predictions
HyperSizer pretest predictions of the MSFC orthogrid 44" square test panel were completed July 2007. The predicted failure load and failure modes are summarized in the table. The failure loads are reported in terms of acreage unit load in (lbs/in). The overall panel failure load was predicted to be 297,000 lbs and included approximate effects of the thickened panel edges.
The unit failure load of the panel was (5300 lb/in), slightly higher than the HyperSizer pre-test prediction of 5035 (lb/in).
Local Skin Pocket Buckling Verification to FEA
HyperSizer local buckling analysis of the skin between stiffeners of an orthogrid cylindrical panel is compared to FEA for uniaxial compression loading. The local buckling images and summary table represent a small portion of the MSFC test panel (3 pockets by 3 pockets) and use the same physical dimesnsions (height, spacing, thicknesses). For the test panel, the maximum difference is less than 2% between FEA and HyperSizer skin local pocket buckling.
An extensive verification effort was undertaken for orthogrid and local buckling of skin pockets, including tension field stiffening effects to a variation of tank pressures - as done for the panel buckling shown above.
Cylindrical Panel Buckling Verification to FEA
HyperSizer cylindrical panel buckling analysis was compared to FEA for different internal tank pressures ranging from zero to 50 psi. The average difference between HyperSizer and FEA predictions was 1%. In all cases, the maximum difference was 3%. The cross sectional dimensions from the MSFC test panel (see below) were used as the example orthogrid design. For FEA verification, the cross section was discretely meshed, using HyperFEMgen, with shell elements representing the skin and webs.
The above image is an FEMs generated by HyperFEMgen and analyzed with different loadings to generate the graphs below of axial compressive buckling load versus internal tank pressure.
The left image is the buckling mode shape with zero pressure. The right image is the buckling mode shape with high pressure, displaying the effect caused by tension field hoop stiffening on mode shapes.
The above graph plots failure load prediction as a function of pressure up to 50 psi. Internal pressure causes an increase in axial compressive buckling load due to hoop tension field stiffening. This effect is represented with the set of curves labeled "No axial pressure relief". This hoop tension field stiffening effect is quantified as an intrinsic part of the buckling solution. An additional benefit is the linear superposition of axial tension load (axial pressure relief) which reduces the applied compressive load. The total benefit of pressure stabilazation is plotted with a set of curves labeled "Including axial pressure relief."
The graph below zooms in on the range of zero pressure up to 10 psi. The comparisons of the four analyses are so close, the green HyperSizer line is covered.
There are several observations to make. First is that HyperSizer's buckling method matches FEA solutions for the orthogrid panel, especially at zero or low internal pressures. Second, HyperSizer generated equivalent stiffness terms for 2D planar finite element meshes with NEi are correct based on comparison to the discretely meshed 3D models. Third, that the discretely meshed 3D FEA solutions between Nx and NEi Nastran are the same. Results and explanations are included in the referenced PPT.
Related Resources
Esoteric Capabilities
Below are listed new capabilities programmed into HyperSizer specifically in support of the Ares upper stage project.
- NASA SP-8007 Cylindrical Panel Buckling Method: For simply supported, full cylinders (e.g. fuselage or cylindrical tanks), this method compares very closely, but has the advantage over HyperSizer’s built-in numerical buckling solution of being very efficient and therefore greatly speeding up optimizations that are controlled by global panel buckling. (Ref: NASA SP-8007, “Buckling of Thin-Walled Circular Cylinders,” Section 4.3.)
- Weight of orthogrid fillet radius, corner radius, and lightening hole: For orthogrid stiffened panels, HyperSizer includes additional weights from fillet radius and corner radius. HyperSizer includes subtracted weight from lightening hole (a hole drilled at the intersection of webs). HyperSizer checks that a lightening hole is not larger than physically possible
- Separate buckling knockdown factors for panel and local buckling: For all stiffened panel concepts, HyperSizer permits the user to enter separate buckling knockdown factors for local buckling and panel buckling. Enter this value on the Buckling tab.
- Local pressure bending effects for skin pocket of an orthogrid stiffened panel: For typically sized orthogrid panels, a large percentage of applied pressure is reacted in hoop force in the skin, with a smaller portion reacted by the stiffeners. Based on the ratio of skin to circumferential web stiffness, some of the internal pressure causes localized bending moments into the skin pocket. This bending stiffness effect is now accounted for as a reduced pressure in the local pressure bending calculation. This is a significant and novel analytical method developed for the hydrogen tank. This effect, and the accuracy of HyperSizer compuations to quantify it, have been verified with detailed FEA.
- HyperFEA automated iteration between HyperSizer and Eigenvalue Buckling FEA solutions: HyperFEA had been capable of controlling the execution of, and data transfer between, HyperSizer and FEA linear elastic solutions. For the NASA Marshall CLV components, in particualar, the common bulkhead, a new capability was developed for also controlling the execution of FEA eigenvalue buckling solutions. As shown above, HyperSizer has accurate methods for predicting panel buckling for cylindrical structures. However, for doubly curved structures such as the common bulkhead, HyperSizer does not contain explicit methods for buckling. In this case, the recommended approach would be to find suitable lengths for buckling spans such that the cylindrical buckling solutions would represent the buckling of a dome fairly well. These types of applications require HyperSizer buckling lengths to be tuned to the specific problem by adjusting the buckling lengths in an iterative manner such that values predicted by HyperSizer match FEA solutions. This process, though proven to be suitable, is time consuming. The new HyperFEA automates this process by:
- Submitting the FEA eigenvalue buckling solution
- Reading the eigenvalues from the computed FEA output
- Defining the ratio of HyperSizer vs. FEA eigenvalues and adjusting the current state of HyperSizer buckling lenghts by this ratio
- Repeating this process while sizing to minimum weight
Frequently Asked Questions
Below are responses to frequently asked questions from MSFC users.
- How Do I Analyze and Size Orthogrids to Panel and Local Buckling Using Only Internal Tank Pressure Relief and Not the Added Benefit of Hoop Tension Field Stiffening?
On the HyperSizer FBD tab, the loads entered into the row labeled “For strength analysis” are used for material stress, local buckling, and crippling. The loads entered into the row labeled “For buckling analysis” are used for panel buckling.
- How Do I Set Fabrication Limits on Orthogrid Panel Construction and Sizing?
Due to the bump forming process of the Al-Li orthogrid panels, limits are set on the Web dimensions. The current limits, and the way users can enter these values in the Backdoor form entry, is shown directly below.
- What is Backdoor Data?
HyperSizer project Backdoor GUI data entry is a temporary way for HyperSizer developers to provide quick turn around new capabilities to MSFC end users. Eventually, some of the capabilities initially available through this rudimentary GUI input get relocated to the primary GUI. For example, local buckling knockdown factor and fillet and corner radius values are input that migrated from the HBD to the interface.
More Information: