King of Lightweight Design


General Dynamics

Convair Aerospace Div.

San Diego, Calif.

What’s the most efficient structural design? Today’s front-runner is a super-light panel based on triangular reinforcement elements. Although it was conceived for aerospace applications, there’s no reason why the idea can’t be put to work in commercial fields as well.

THE SEARCH FOR LIGHTWEIGHT STRUCTURES will never end. One of the latest in a long legacy of lightweight design is the isogrid panel--a one-piece structure in which a network of load-bearing ribs and a protective skin are whittled out of a solid aluminum plate. Born in spacecraft design, nurtured by aerospace development, lightweight isogrid panels are ready for some earthbound applications.

The earliest lightweight structures were skin-stringers dating back to dirigibles and the early days of aircraft. Basically, stringers, often wood or aluminum ribs, form the panel skeleton. A skin of fabric, or aluminum sheet, stretches across the skeleton and encloses the panel.

With modern material and manufacturing improvements, machining an isogrid network from a solid plate of metal is an economical alternative to the skin-stringer. Isogrid panels can be 20 to 40% more efficient than comparable skin stringer structures, and can be made for about 50% of the cost. The one-piece isogrid eliminates fasteners and fastener beef-up areas. Fewer parts and reduced assembly time mean lower cost.


Fig. 1--The rectangular, or waffle pattern isogrid. This integrally machined panel was developed eight to ten years ago at General Dynamics.


So What’s New?

The integral, machined isogrid is not new. However, the demands for light weight and high strength haveinspired some appealing design refinements. Isogrid ribs can assume any geometric pattern. The rectangular, or waffle pattern, originally developed eight or ten years ago, is used occasionally. But rectangles, Fig. 1, are inefficient because only ribs in the direction of loading are reacting. A triangular isogrid, in contrast, distributes the load among all ribs. With every member doing its share of the work, dead weight is virtually eliminated. Equilateral triangles, Fig. 2, are usually preferred because they require minimum tool change during machining. (The same tool can machine each corner of the triangle.)


Fig. 2--Triangular isogrids make most efficient use of all structural members. All ribs share the applied load.



Simple triangular isogrids can be made with straight-sided ribs, Fig. 3. However, by undercutting the ribs to form an I-beam cross section, Fig. 4, panel weight can be reduced 10 to 20% without loss of strength. The increased stability and load-carrying capacity of the I-beam or "flanged" ribs, as contrasted to straight or "unflanged" ribs, make the flanged isogrid panel the most efficient lightweight load-bearing structure to date.



Fig. 3--Straight-sided isogrid. Skin helps support the load in this design.

Skin-stringers and flanged isogrid structures both tend to react in the same way under loading. In tension, the ribs and skin share the load. In compression, the skin takes only a fraction of the load, leaving the burden on the ribs. In unflanged iso-grid design, however, the skin must be thick enough to assume as much compressive loading as possible. For maximum efficiency, every component of the structurecontributes to the overall strength. When the critical compressive strength of the skin is exceeded, the skinbuckles and transmits severe, uneven strains to the ribs, forcing them to buckle. The panel collapses.



Fig. 4--Flanged isogrid. The I-beam cross section of the ribs is strong. Skin is allowed to buckle elastically under compressive loading.



Fig. 5--NC milling machine whittles away at an aluminum plate at General Dynamics. Skin is milled to 0.040 in. thick.


Fig. 6--NC milling machine, set up with a T-slot cutter, forms the 1-beam cross section in this flanged iso-grid.


The flanged isogrid, however, does not follow this sequence of failure. Proportionally, the flanged isogr’id is so strong that the skin’s contributions to the overall panel strength is limited to providing grid-member column stability in the plane of the skin. In fact, the unique aspect of flanged isogrid design is that the skin thickness is reduced to a point where it buckles elastically under light compressive loading. (This buckling does not affect the flanged rib stability. A relatively thin skin transmits less strain to the ribs and is less likely to cause complete panel failure.)

With the skin buckled and out of the way, the strong flanged ribs carry the compressive load unimpeded. As the load is removed, the skin returns to its normal state. The result: even greater weight reduction for the flanged isogrid because of the thinner skin.

Under tension, the reaction to loading is slightly different. The skin develops a tension field under tensileloading, but still assumes some of the tensile load.

Making Isogrids

Isogrid panels are easily machined from solid plates by numerical control milling, Fig. 5. To simplify NC

programming, the panels are machined flat and bent to shape after milling. The isogrid pattern is rough and finish end-milled to the I-beam flange width. The web of the I-beam is formed with a T-slot milling cutter, as shown in Fig. 6.

It is not practical to end-mill skin thickness under about 0.040 in. For thinner sections, tool forces lift skinsections off the chuck, resulting in punctures and irregular cross sections. Skin may be thinned below 0.040 in. by chemical milling. Webs, flanges and other features must be protected by masking during chemical milling.

The machined isogrid panel can be rolled or brake formed after the machined-out pockets are filled with alow-melting epoxy filler, Fig. 7. The filler stabilizes the grids and skin, evenly distributing the forming forcesthroughout the structure. Forming without a filler may result in irregular skin and occasional rib failures. Thefiller is melted out after forming.

Manual bending and brake bending rough-shape the panel, Fig. 8. For more accurate sizing, the panel is clamped to a forming fixture and aged, Fig. 9.

Although the procedure for manufacturing iso-grid panels seems long and tedious, cost studies indicate a 64% reduction in recurring costs and 37% reduction in non-recurring costs over equivalent structures made from skin-stringers.

Fig. 7--Machined panels at General Dynamics are filled with a low-melting resin to allow easier bending and forming.



Fig. 8--Forming the isogrid. This 3/4 in. thick panel is rolled manually to accuracies of ± 0.10 in. at General Dynamics.



Fig. 9--Final sizing the isogrid. Manually-formed panel is clamped to a forming fixture and aged by General  Dynamics, to achieve more accurate shape.

Isogrid Geometry

The first consideration when laying out an iso-grid is "greatest possible strength with least possible weight." Basically, isogrid design involves a trade-off between triangle-node spacing and I-beam

cross section. For panels bolted together, the node-to-node spacing should be an integer multiple of the bolt spacing. The grid pattern blends smoothly into transition points where panels are joined together, Fig. 10. Ideally, openings and other discontinuities in the panel follow the grid configuration. For example, grids blend smoothly around ports in the panel, Fig. 11.





Fig. 10--Two panels spliced together. Grids are laid out to permit smooth transitions at bolted joints.




Fig11--Access openings in isogrid panel. Grids are layed out to follow openings and discontinuities.

Node-to-node spacing should be as large as possible for minimum weight, but this spacing is limited by structural loading requirements. The I-beam supports the panel load, so its cross section is determined by peak edge-load intensity. Isogrid geometry is usually expressed as the ratio of node-to-node spacing to I-beam height. Typical ratios range from 9/1 to 16/1. Most often, some ratio within this range provides the best grid layout, with smooth transitions--and openings if required--and an adequate 1-beam cross section at lowest possible weight.

When an isogrid panel is bent or formed, I-beam height is also restricted by the forming radius. The allowable forming radius for any I-beam cross section is limited by the yield point of the material. Bending has been satisfactory when the I-beam height results in no more than 2% elongation after forming. Greater elongation places the skin in a state of tension and may induce premature failure. When severe elongation is unavoidable, increase skin thickness to allow for stretching and thinning. Excess material can be removed after forming by chemical milling.





Fig. 12--Section of an aerospace adapter made by General Dynamics from three spliced isogrid panels.

To What End?

Isogrids are now found only in experimental aircraft and aerospace vehicles. Formed and assembled into large, cylindrical structures, entire sections of spacecraft are made of these panels, Fig. 12. Structural tests, soon to be conducted at NASA’s Marshall Space Flight Center, will be aimed at further evaluating the potentials of the isogrid structure. But even other uses are being considered. With the high strength-to-weight ratio and exceptional stress distribution under load, isogrids are candidates for:

s Ribs of a parabolic or paraboloidal dish type reflector for radio frequency antennas, solar collectors, or

telescopes. The exact parabolic contour can be machined into the ribs more efficiently than by conventional

forming methods.

s Any lightweight, portable pedestal or enclosure required to carry high loads. Portable military towers, bridges, scaffolding, and shielded r-f enclosures are a few examples.

s Manually movable decorative walls and partitions.

s Flight-weight or lightweight storage tanks that must retain good structural characteristics when non-pressurized. Also tanks that must carry a body-bending or axial load that pressure stabilized skin cannot handle.

s Future aircraft, particularly hypersonic aircraft.

s Flight-weight dewar structures.

s Impact absorbent structures. Because of the excellent load redistribution in isogrids, there is no well-defined failure load.