Tools › Structures › Composites › Sandwich Panel Inserts › Method

Sandwich Panel Inserts: Method

Open the Sandwich Panel Inserts tool

Sandwich Panel Inserts: Core Shear Around the Potting, Partial Potting, Potting Rupture and Combined Loads

The insert and the question

Equipment is fastened to a honeycomb sandwich panel through inserts: threaded metal bodies set in a bore in the core and bonded there with a potting compound that fills the opened cells. A load normal to the panel can only leave the insert as shear in the core around the potting, helped a little by the bending of the face sheets, so the pull-out strength is set by the thin cell walls and not by the insert. The tool estimates the static strength of one insert for a normal load, an in-plane load, a bending moment and a torsion, and the margin of the combination.

The method is the analytical model of the ESA Insert Design Handbook, ECSS-E-HB-32-22A (2011), which continues ESA's earlier insert handbook and rests on tests of several hundred inserts in aluminum cores. Equation numbers in square brackets are the handbook's. The handbook is cited, not reproduced: none of its property tables or capability charts is on this page, and every material value is an input.

Three kinds of insert

A through-the-thickness insert spans the core and is bonded to both face sheets. A fully potted insert is set from one side with potting that reaches the far face sheet. A partially potted insert is set from one side in a core deeper than the potting, so that unfilled core lies below it. The handbook treats an insert as fully potted when the core height c is not more than the insert height plus 7 mm, hi + 7 mm [7.5-1]: the bore is 3 to 4 mm deeper than the insert and the resin must connect at least 3 mm of core below it. For a deeper core the minimum potting height is hp = hi + 7 mm [7.5-2] and the typical one hp = hmin + A tanh((c − hmin) / hmin) [7.5-3], with A = 2.5 mm for 3.2 mm cells and 5.0 mm for 4.8 mm cells.

Potting radii

Two radii describe the potting. The effective potting radius bp is the mean distance from the insert center to the nearest single cell walls around the potting; the double walls next to the resin are stiff and are counted as part of it. It carries the core shear. The real potting radius bR is the radius of a circle with the area of the resin and carries the tension in the potting and the torsion. With bi the insert radius and Sc the cell size, in millimeters:

perforated core (aluminum): bp,min = 0.93192 bi + 0.874 Sc − 0.66151 [7.3-2], bp,typ = 1.002064 bi + 0.940375 Sc − 0.7113 [7.3-4];

non-perforated core: bp,min = 0.9 bi + 0.7 Sc [7.3-3], bp,typ = bi + 0.8 Sc [7.3-5];

real radius: bR,min = bi + 0.35 Sc [7.4-3], bR,typ = bi + 0.5 Sc [7.4-4].

Minimum values are for design, typical values for comparison with tests. Measured radii can be entered instead.

Core shear around a rigid insert

A circular sandwich plate loaded by a force P through a rigid central insert of radius b was solved by Ericksen and by Youngquist and Kuenzi at the U.S. Forest Products Laboratory, and the solution is the basis of MIL-HDBK-23A chapter 20 as well as of the ESA handbook. Far from the insert the core carries the whole load and its shear stress falls as 1/r. Next to the insert the face sheets, bending about their own mid-planes, take part of the load, so the core shear stress is zero at the edge of the potting, rises to a peak a short distance away and then follows the 1/r curve. In the handbook's notation [C.2-1], [C.2-12]

τ(r) = (P / (2π bp c)) C* K(r), K(r) = bp (1/r − A I1(αr) − B K1(αr)),

where I1 and K1 are modified Bessel functions, the constants A and B make K vanish at the insert and satisfy the condition at the panel radius a, and

α = √(Gc (h − c) I / (E′ c f1 f2 Is)), E′ = Ef / (1 − νf²), Ip = f1 f2 (h + c)² / (4 (h − c)), Is = (f1³ + f2³) / 12, I = Ip + Is, C* = (Ip / I)(c / d).

Here h = c + f1 + f2 is the panel thickness and d the distance between the face sheet mid-planes. For equal face sheets C* = β(β + 1) / (β² + 2β + 4/3) with β = c / f [C.2-8], which the handbook rounds to β / (β + 1) [C.2-14]; the tool keeps the unrounded form and accepts unequal thicknesses. For a large panel the solution reduces to K(r) = bp/r − K1(αr) / K1(αbp).

The handbook replaces the Bessel functions by exponentials, K = (bp/r)(1 − √(r/bp) eα(bp − r)) [C.2-6], and gives the radius of the peak as r = bp / (1 − exp(−0.931714 (αbp)0.262866)) [C.2-15]. The insert load at which the peak reaches the core shear strength is

Pcrit = 2π bp c τc / (C* Kmax) [C.2-17].

The exponential form is close to the Bessel solution when αbp is above about 5 (0.1 % at αbp = 23, 1 % at 10; the fitted radius drifts again for very large values, 0.5 % at 400) and falls below it for small values (4 % at 5, 13 % at 2), which overstates the capability; that happens with thick, stiff face sheets on a soft core. For that reason the tool uses the Bessel solution for a large panel by default; the exponential form can be selected in Advanced mode, with a warning when αbp is below 5, and the page prints Kmax both ways. The stiffness ratio of MIL-HDBK-23A, φ = √(h Gc / DF) with h the distance between facing centroids, differs from α by the factor √(d I / (c Ip)), a few percent; with φ the Bessel solution reproduces the handbook's worked example (k3 = 0.54 at b/a = 0.091, φa = 40) and its chart for φa of 8 and more.

Effective core properties. Around a circular potting both foil directions carry shear, and the single foils buckle early. The handbook therefore uses τc = 1.36 τW [6.5-1] and Gc = GW / 3 [6.4-1], from the W-direction strength and modulus of the core. Both factors are inputs in Advanced mode.

Partial potting

Metallic core. Below a partial potting the load divides into a part carried by the loaded face sheet, a part carried as shear over the potting height and a part carried as tension (or compression) in the core under the potting. Because the foils buckle and yield, the handbook takes the core shear strength and the core flatwise strength to be reached together [C.3-10]:

Pp,crit = ½ Pcrit + π r (2 hp − c) τc + π r² σc,

with r the radius of the peak shear stress and σc the flatwise tensile strength of the core for a pull-out load, the compressive strength for a push-in load. The lower of Pcrit and Pp,crit applies: in a shallow core the insert behaves as fully potted although it is not.

Non-metallic core. Stiff aramid or glass cell walls cannot spread the stress concentration at the bottom of the potting. The handbook covers it by P = Pcrit (hp / c)0.62 [C.3-11], [C.3-13], and rates the result as suitable for preliminary design only.

Rupture of the potting below the insert

Under a pull-out load the resin under the insert is in tension. With the load divided as before but the shear taken over the insert height hi, and the tension in the resin spread over the real potting area, the insert load at which the resin reaches its tensile strength σR in a metallic core is [C.4-11], written in the form of the handbook's C.4.3,

PR,crit = π bR² σR / (½ + (C* Kmax / (bp c)) r (c/2 − hi)),

and in a non-metallic core P = π bR² σR [C.4-13]. The strength to use is the one effective in an insert, which includes the stress concentration at the insert flange and is lower than the strength of a cast specimen. A taller insert is the remedy when this mode governs.

Push-in loads and thick face sheets

Under a push-in load the potting is not critical. With a thick face sheet the bond between the face sheet and the core can fail in tension beside the insert, so the face sheet share is left out: P = ½ Pcrit + π r c τc for a fully potted insert [C.5-1], 2π r hp τc + π r² σc for a partially potted one in a metallic core [C.5-2], and the fully potted value times (hp / c)0.62 in a non-metallic core [C.5-3]. The handbook's chapter 13 applies this to face sheets thicker than 0.6 mm; that is the default rule here and can be overridden.

In-plane load, bending and torsion

In-plane load. A metallic face sheet fails by compressive buckling ahead of the insert, limited by its yield strength σf; the clamped core adds a small part. The semi-empirical capability is QSS = 8 bp² τW + 2 f bp σf [14.2-1], with bp not more than 11 mm in this relation and with the foot of the attached part at least as wide as the potting. For a through-the-thickness insert each face sheet carries 2 f bp σf [14.3-2] and the core is given no share.

Bending. MSS = PSS,c bi [15.2-1], the compressive capability acting at the insert radius. A through-the-thickness insert reacts a moment as a couple of the two face sheet shear loads, M = d × (face sheet shear capability). The handbook advises against bending a single insert.

Torsion. In a metallic core the cell walls enclosing the potting shear: TSS = 4π bR² t0 τ0 [16.2-1], with t0 the foil thickness and τ0 the shear strength of the foil material. For non-metallic cores the handbook asks for a test; the value is an input.

Edges and neighbors

Near a free panel edge at distance e the capability is multiplied by ηEN = 0.55 √(e / bp) − 0.05 e / bp for a normal load when e ≤ 5 bp [18.1-2], and by ηEQ = 0.66 √(e / bp) − 0.06 e / bp for an in-plane load when e ≤ 3 bp (metallic face sheets) [18.2-2]. An edge with a close-out member needs no reduction: leave the edge distance blank. A neighboring insert at center distance a ≤ 5 (bp1 + bp2), loaded in the same direction up to its own capability, reduces the normal-load capability by ηIS = (ρ / (1 + ρ)) (1 + (a / (5 bp1)) / (1 + ρ)), ρ = bp1 / bp2 [19.1-3]. Both relations step slightly at their limits (0.98 and 0.96 instead of 1 for the edge coefficients).

Combined loads and margins

The handbook's interaction is quadratic [17.2-1]: (P / PSS)² + (Q / QSS)² + (M / MSS)² + (T / TSS)² ≤ 1. With Rc the square root of the sum, all loads can be raised together by the factor 1 / Rc, so the margin of the combination is 1 / Rc − 1. For a through-the-thickness insert the in-plane load and the moment are first resolved into the shear of one face sheet, Q / n + M / d, and that ratio replaces the two terms (handbook 17.2).

Every check is a strength check, so the margins are written at ultimate load: the entered loads times the factor of safety when limit loads are entered. The handbook finds that significant deformation of an insert begins close to its ultimate load and asks for no separate yield check (12.4.2). The knock-down factor multiplies every capability. The handbook's own treatment multiplies its minimum values by a reliability coefficient that depends on the core, the core height and the face sheet thickness (its 12.3 and Table C-2) and then allows a factor of safety of 1.5; that coefficient is not built into this page, enter it or a program value as the knock-down factor.

What it assumes

A single insert, rigid compared with the panel and clamped by the attached part, in a flat panel large compared with the potting (panel radius more than twice the potting radius). Honeycomb core with negligible in-plane stiffness and face sheets of one isotropic material; a composite face sheet enters through √(Ex Ey) and √(νxy νyx), as the handbook does. Static load at room temperature. The potting is sound and at least as large as the relations assume. The strengths entered are the ones wanted for the purpose: minimum values with the minimum potting radius for design, typical values with the typical radius to predict a test.

When it misleads

The model was fitted to aluminum cores with thin aluminum face sheets. Partial potting in non-metallic cores, composite face sheets under in-plane load, bending and torsion have little test backing in the handbook, which says so; for composite face sheets with ±45° plies its in-plane relation correlated poorly with tests. Core heights above about 80 mm and face sheets above about 0.8 mm are outside the range of the handbook's correlation. The exponential approximation of Kmax, if selected, overstates the capability when αbp is small. The flatwise tensile strength of a core and the effective tensile strength of a potting compound are rarely tabulated: the defaults for them are illustrations. Potting quality dominates the scatter of real inserts, and a void under the insert defeats every relation here. Fatigue, thermal cycling, elevated temperature, the thread and body of the insert and the fastener are not checked.

Checks made on this tool

The Bessel-function solution against MIL-HDBK-23A chapter 20 (example 1 and chart readings); the handbook's printed potting radii, its in-plane example of F.12 (1,589 N and 1,882 N), edge and neighbor coefficients of the same example, the core shear value of its Table 12-4 (3,784 N, reproduced with an assumed aluminum face sheet modulus that the table does not state) and the potting rupture value of its Table B-2 (1,896 N, with its reliability coefficient); and an independent numpy and scipy calculation of sixty cases. The other two values of Table B-2 and Table 12-5 could not be reproduced from the data printed with them.

Material values

Default core: 5052 aluminum honeycomb, 1/8 in cells, 0.001 in perforated foil, 4.5 lb/ft³, with the average W-direction shear strength (230 psi), shear modulus (29.4 ksi) and flatwise compressive strength (580 psi) of MIL-HDBK-23A Tables 2-1 and 2-2; averages, not design minima. Flatwise tensile strength 1,050 psi: an estimate by the handbook's rule (foil tensile strength times core density over foil density) from the 5052-H38 values of MIL-HDBK-5J, not a test value. Face sheets: 2024-T3 sheet, MIL-HDBK-5J Table 3.2.3.0(b). Foil shear strength: 5052-H38 sheet, MIL-HDBK-5J Table 3.5.1.0(b). Potting tensile strength 1,500 psi: an illustrative round number without a source; it governs the default case (potting rupture sets the default pull-out capability).