There is a family of printed parts that almost nobody designs on purpose. It includes the clamp that holds a phone to a monitor arm, the clip that keeps a bag of coffee shut, the hook that holds a bike in a garage, the handle on a drawer, the bracket that keeps a webcam pointed at a desk and the little jaw that grips a shelf edge. All of them exist to hold on to something that was not designed to be held, and all of them fail the same way: they let go. Not with a bang, but with a slow creep and one morning the hook is on the floor and the bike is leaning against the wall.
This article is about the load logic underneath that family of parts. Not the aesthetics, not the infill pattern, not the question of which filament to buy. The question is narrower and more useful: if you have a printed part whose whole job is to grip something, what geometry decides whether it grips or lets go, and what numbers can you sketch on a napkin before you open the slicer?
It connects directly to three earlier pieces in this series. The part your hand actually touches covers the contact surface and the shape of fingers. The brackets, holders and adapters article covers the other end of the problem, where the screw hole cracks. And creep, fatigue and preload covers the time axis. What has been missing is the middle: the beam, the jaw, the hook and the friction that live between the thing being held and the thing doing the holding.
Four jobs that look like one job
Clamps, clips, hooks and handles are usually discussed as if they were variations on a theme. They are not. They are four different load cases that happen to share a visual language, and the design rule that saves one will ruin another.
| Part | What it really does | Dominant load | Failure mode |
|---|---|---|---|
| Clamp | Closes a gap and squeezes something between two faces | Bending in the arm, compression at the jaw | Arm deflects, jaw opens under load, slip |
| Clip | Deflects elastically to snap over an edge, then holds by spring force | Bending strain in the flexure | Yield on installation, or relax and lose grip |
| Hook | Carries a hanging load from a single point | Bending at the root, peeling at the attachment | Snaps at the throat, or the root peels off its fasteners |
| Handle | Transfers a pull force from a human hand to a panel | Bending plus bolt tension plus contact pressure | Fixing holes ovalise, or the grip creeps loose over months |
Read the table again and notice that only one of these four is primarily a friction problem. A clamp holds by friction and by interference. A clip holds entirely by elastic preload, which is a form of stored spring force that decays. A hook holds by geometry and does not care about friction at all, but its root is in bending and its fasteners are in peeling. A handle holds by bolt tension, which is the one thing plastic is worst at maintaining.
The most common design error in this family is applying a clamp mentality to a hook. A hook with a thick, chunky, generous profile feels strong. It is usually weak, because material has been added where the bending moment is lowest and the throat, which is where the moment is highest, has been left thin to preserve the opening. A hook is a cantilever beam with a curve in it, and curves in cantilever beams are where stress concentrates.
Clamps: a beam, a jaw and a screw that never lies
Every clamp is a C-shape carrying its load across an open mouth. That geometry is a bending problem with a very predictable answer, and the answer is more brutal than most people expect because deflection scales with the cube of span.
The classic formula for the deflection of a cantilever or a simple beam under a point load, written in the form that matters at a desk, is:
deflection is proportional to load times span cubed, divided by stiffness times second moment of area.
The practical consequence is that doubling the jaw depth of a clamp makes it eight times more flexible if nothing else changes. Not twice. Eight times. A clamp that grips a 30 mm shelf edge is a very different structure from a clamp that grips a 60 mm shelf edge, and the same drawing scaled up will not work.
| Jaw span | Relative deflection, same load and section | What it means in the workshop |
|---|---|---|
| 20 mm | 1.0 | Open jaw works fine at light load |
| 40 mm | 8.0 | Jaw visibly flexes; add depth to the back |
| 60 mm | 27.0 | Needs a closed or ribbed back to survive |
| 80 mm | 64.0 | Treat it as a frame, not a clamp |
The way out of this is not to make the arm thicker. Making the arm thicker helps linearly with the second moment of area while adding mass and print time. The way out is to close the section: add a back web, a rib along the outside of the C, or a flange. A 2 mm wall doubled to 4 mm gives four times the bending resistance for a doubling of cross section. Two 2 mm walls separated by 10 mm gives roughly the same four times for a fraction of the material, because the material is now working at a distance from the neutral axis instead of sitting on it.
Where the jaw actually slips
A clamp holds something by friction, and friction is normal force times a friction coefficient. Printed plastic on painted steel is somewhere around 0.2 to 0.4. Printed plastic on smooth aluminium is often worse, and printed plastic on itself is around 0.3. These numbers are genuinely low, which is why a clamp that feels tight can still slide sideways.
The normal force from the screw goes straight into contact pressure, and the surface area of the jaw matters less than people think. Friction does not care about area, so enlarging the jaw pad does not increase grip. What enlarging the pad does is reduce contact pressure so the part stops marking the surface and stops crushing. Two different jobs, two different geometry decisions.
Three fixes are worth knowing:
- A rubber pad raises the friction coefficient to around 0.7 and beyond, which is a bigger improvement than any geometry change. A strip of TPU or a silicone sheet costs almost nothing and more than doubles grip. Print the pads in 95A TPU for jaw faces, or in 85A TPU where you want a softer, more conforming contact.
- A mechanical stop is better than friction when you can arrange it. A small lip at the bottom of the jaw that the clamped object rests on converts a friction problem into an interlock. Nothing slides downhill past a lip.
- Angle the jaw faces so that the load pulls the clamp tighter rather than looser. A two degree negative rake towards the back of the jaw makes self-tightening behaviour free.
Do not clamp with a printed screw
A printed thumb screw threaded into printed plastic is a mechanism with two compliance sources in series. The thread is loose enough to rock, the plastic creeps under thread pressure, and the knurl that makes the screw turnable is usually the first thing to shear off. If the clamp needs a screw, use a metal bolt into a heat-set insert, which has its own article on geometry and depth, or use a lever, or use a cam built on a flexible arm. A cam clamp printed as one piece with a flexible hinge is one of the few mechanisms where printed plastic is genuinely the best material, because it exploits compliance instead of fighting it.
Clips: the flexure is the feature
A clip is a beam that must bend far enough to pass over an edge and then spring back to hold. Everything about a clip follows from one number: the strain at the surface of the flexure at maximum deflection. If that strain exceeds the material's yield strain, the clip bends permanently on installation and never grips again. It is not slightly weaker. It is a different part with a different job.
Printed filament plastic yields in the low single-digit percentages. PLA is brittle and gives up early, around three percent for most blends. PETG is closer to five or six in bending and will complain loudly by whitening first. Nylon and the polyamides are more forgiving still, and TPU basically does not care, which is why TPU is the default for clips whose only requirement is to grip.
The strain in a bending beam is a simple ratio: thickness of the beam divided by twice the bend radius. That ratio is the single most useful equation in this whole family of parts, because it tells you immediately that a thin long beam is compliant and a thick short beam is not, and it removes the temptation to make clips chunky.
| Flexure thickness | Minimum bend radius for 3% strain | Practical reading |
|---|---|---|
| 1.0 mm | 17 mm | Compliant, works in PLA |
| 1.5 mm | 25 mm | Compliant, works in PETG |
| 2.0 mm | 33 mm | Marginal in PLA, fine in nylon |
| 3.0 mm | 50 mm | Not a clip, a machine element |
Two consequences follow that are worth stating plainly. First, the flexure thickness must be a printing dimension you can actually hold. A 1 mm flexure printed with a 0.4 mm nozzle is two lines with a seam, and one of those lines can land on a slightly colder layer. A flexure printed as an odd number of perimeters with a defined wall order is more reliable than one defined only by an STL thickness. Second, the direction of the flexure relative to the layer lines decides whether it survives. A beam bending across the layer lines is loading the weakest bond in the part in its worst possible mode. A beam bending within the plane of the layers is roughly ten times better. Any clip whose flexure is oriented so that its bending axis lies along the print direction is a clip that needs no other design work.
The spring that forgets it is a spring
A clip holds by stored elastic force, and stored elastic force decays. Printed plastics under continuous strain relax over days and weeks, and the amount depends on temperature, load level and material. A clip that grips a 3 mm panel edge on day one with a satisfying snap can be loose by month two, not because it yielded but because it relaxed.
The design response is to give the clip a defined snap engagement rather than a friction grip. A raised bump that catches behind the panel edge is a geometric hold, and it does not care about preload decay. The spring force then only has to be enough to reach that bump and keep the clip from rattling, which is a much easier requirement than retaining the part by friction alone. This is the same design move that makes snap fits and detachable joints reliable, and the same reasoning that makes compliant mechanisms work without hinges.
Hooks: the throat is the part
A hook has one characteristic dimension, which is the radius of the throat or the material thickness at the innermost point of the curve. Everything else about the hook is secondary, because the bending moment from the hanging load is greatest at the throat and drops towards the tip.
If you hang a load of mass m at the tip of a hook with a horizontal reach of L, the bending moment at the throat is roughly m times gravity times L. For a 5 kg load on a hook reaching 40 mm from the throat, that is about 2 N·m. Printed plastic in a beam of 10 mm width needs a depth that gets the section modulus above the moment divided by allowable stress, and the answer for most filaments at a reasonable safety factor comes out in the range of 8 to 12 mm of material at the throat.
| Hanging load | Reach to throat | Moment at throat | Suggested throat depth, 10 mm wide, PETG |
|---|---|---|---|
| 1 kg | 30 mm | 0.3 N·m | 5 mm |
| 3 kg | 30 mm | 0.9 N·m | 8 mm |
| 5 kg | 40 mm | 2.0 N·m | 11 mm |
| 10 kg | 40 mm | 3.9 N·m | 15 mm |
| 20 kg | 50 mm | 9.8 N·m | Use steel, not plastic |
These are design starting points with a generous safety factor, not a licence to hang 20 kg on a printed hook. The load ratings that matter are the ones imposed by the material's long-term creep, not its short-term strength, which is why a hook that holds for an hour and a hook that holds for a year are different parts. A printed hook carrying a permanent load at half its short-term capacity is a hook that will open its own throat over a season and drop the load.
The root is a peeling problem, not a strength problem
Hooks usually break at the base, and they break at the base because the base is where the fasteners are. A hook bolted to a wall has one or two screws, and the load hangs out at a distance from those screws. That is a lever, and the lever works on the top screw as a prying force. Printed plastic bolted flat to a wall with the load pulling away from the wall at the top edge is a connection in peeling, and peeling is where printed brackets crack, a mechanism that has its own article on screw holes.
- Put the load line over the fixing, not beside it. A hook whose centre of load is directly below the top screw has almost no prying moment.
- Add a lip or a heel at the bottom of the mounting plate that bears on the wall. The compression at that heel cancels most of the peel.
- Use a washered through-bolt with a printed boss in compression, or a metal plate as the structural member with the printed hook as a cover.
Handles: preload is a promise you cannot keep for long
A handle bolted to a drawer or a panel is a set of bolts holding a lever. The bolts are preloaded by tightening, and the preload is what keeps the joint tight under a cycling pull. Plastic does not hold preload well. Under the contact pressure at the bolt head, the material beneath the head creeps, the fastener loses clamp load, and the joint goes from a friction connection to a bearing connection in which every load cycle works directly on the hole.
The standard fix is a metal washer under every bolt head, giving the bolt a hard surface to bear on. Most printed handle failures come from skipping this step; it is much harder to recover once the hole has ovalised. The second fix is deeper: put a printed boss under each bolt so the washer has a real seat instead of a curved surface. The third is to make the handle as a printed shell with metal bolts, with the boss in pure compression and a metal plate spreading load; the plastic then carries no structural duty.
Handle geometry has one more trap that is specific to this family. The natural design instinct is a solid bar with fillets, which is heavy, slow to print and stiff enough that the hand absorbs all the shock. A hollow handle with a closed section and a rounded underside is lighter, stiffer for the material, and far more comfortable. Print it with the long axis vertical, or as a split shell with wall thickness chosen deliberately and joined on a plane, so that the bending loads are carried within the layers rather than across them.
Friction is a design parameter, not a mistake
Almost all of the parts in this family depend on friction somewhere, and friction is the number designers most often guess. The table below is the range of what printed surfaces actually do, and the practical point is that the spread is wide enough to make the whole design work or fail.
| Contact pair | Approximate dry coefficient | Note |
|---|---|---|
| Rigid filament on painted steel | 0.2 to 0.4 | The optimistic case; paint is not uniform |
| Rigid filament on smooth aluminium | 0.15 to 0.3 | Worse than people expect |
| Rigid filament on glass | 0.2 to 0.3 | Very sensitive to dust and fingerprints |
| TPU on steel | 0.6 to 1.0 | The cheapest upgrade available |
| Printed on printed, same material | 0.25 to 0.35 | Good for nested parts, poor for grip |
| Knurled filament on fabric | 0.5 to 0.8 | Geometry beats surface chemistry |
Two design moves exploit this table. The first is a TPU pad, which is worth more than a doubling of clamp force for a fraction of the complexity. The second is geometry that does not depend on friction: a ribbed or toothed contact surface that interlocks with a compliant material, a lip that catches an edge, or a taper that wedges. A wedge is worth mentioning separately because it is the only pure-geometry solution: a five degree taper pushed home multiplies normal force by roughly eleven times the applied push, which is why tapered printed wedges grip things no flat clamp can hold.
Materials: what each one is actually good at in this family
The material choice for a clamp is different from the choice for a clip, which is different again from a hook. PLA is fine, and even preferable, for any of these parts that live indoors in a controlled temperature and never see a sustained load. It is stiff, it prints accurately, it takes a nice surface, and its brittleness only matters when the part is asked to bend a lot or to take an impact. A PLA handle on a desk drawer is a perfectly good part. A PLA clip on a garage shelf is not.
| Material | Good for | Bad at | Typical use in this family |
|---|---|---|---|
| PLA+ | Stiffness, accuracy, looks | Creep, impact, heat | Indoor handles, display hooks, light clamps |
| PLA Matte | Surface that hides layers | Same limits as PLA, plus slightly lower strength | Visible handles and brackets in a home |
| PETG | Toughness with moderate stiffness | Creep under sustained load, stringing | Clamps, general hooks, shop fixtures |
| ABS | Heat resistance, impact | Warping, fumes, needs an enclosure | Car interior parts, warm-environment clamps |
| ASA | UV and weather resistance | Same printing difficulty as ABS | Outdoor hooks, garden brackets, bike storage |
| ABS-CF | Stiffness plus heat resistance | Brittle at high fibre loading | Stiff clamps that must not deflect |
| PETG-CF | Stiffness in a tough matrix | Abrasive to nozzles, still creeps | Structural clamps and heavy-duty hooks |
| TPU 95A | Grip, damping, tolerance of deflection | Too soft to be a clamp body | Jaw pads, clip bodies, protective faces |
| TPU 85A | Soft conforming contact | Hard to feed, very slow | Pads for irregular or delicate surfaces |
| TPU 95A High Speed | TPU at usable print speed | Still not a structural material | Production runs of clips and pads |
| PEBA | High energy return, spring behaviour | Cost, drying discipline | Clips that must not relax, latches |
| PA6-GF25 | Stiffness, heat, fatigue resistance | Moisture, warp, needs a hardened nozzle | High-load clamps and load-bearing hooks |
| PA6-CF20 | Very high stiffness and strength | Brittle, moisture sensitive | Compact high-load brackets |
| PA612-CF15 | Strength with lower moisture uptake | Cost | Outdoor or humid-environment hooks |
| PET-GF15 | Stiffness with good layer adhesion | Moderate creep resistance only | Structural clamps for shop use |
| PPS-CF10 | Heat and chemical resistance with stiffness | Print temperature, cost, warping | Engine-bay, industrial and hot-environment clamps |
Two notes on the table. Nylon-based materials are the right answer for any clip expected to snap on and off thousands of times, and the wrong answer for anyone unwilling to dry filament properly; wet nylon prints weak and the failure will look like a design error. And the fibre-filled materials are stiffer and stronger but also more brittle in bending, which makes them excellent clamp bodies and poor clip flexures. Fibre reinforcement helps where the load path is in tension and compression, and hurts where the load path requires strain.
Testing a clamp before you trust it
The parts in this family fail slowly, so a five minute test that only proves it holds today tells you nothing. The useful tests are the ones that reveal the time dependence.
- Load it to twice its working load and leave it for an hour. If it creeps visibly in an hour at room temperature it will fail in service, because service is thousands of hours.
- Load it at the temperature it will actually see. A garage in summer is not 20 °C. A clamp that holds at room temperature and lets go at 40 °C is a clamp that works only in winter.
- Cycle the clip fifty times and measure the grip. If the grip force drops noticeably in fifty cycles, it will be worse in five hundred.
- Load it off-axis. Real loads are never perfectly aligned, and the parts in this family are always loaded at a slight angle in use. A hook loaded sideways can fail at a quarter of its rated hanging load.
- Soak it in the environment. A wet bathroom, a kitchen, an outdoor bracket and a garden hook are all accelerated creep tests. If the part is going outdoors, use ASA or a polyamide and expect a shorter life than the specification suggests.
And one measurement worth doing at the drawing stage: put the load path on the sketch. Draw the line of action of the load, then draw the shortest distance from that line to the nearest fixing. That distance multiplied by the load is the prying moment the fasteners must resist, and it is the number that predicts whether the part peels off its mounting rather than breaking in the body. Almost every failure in this family that surprises the designer is a prying moment that was never drawn.
The design checklist for anything that grips
Clamps
- Is the jaw span under 40 mm? If not, add a back web or rib rather than thickening the arm.
- Is the contact face a lip, a pad, or a rake angle? Pure flat-on-flat friction is the weakest option available.
- Does the screw go into metal, or into printed thread? Printed thread is a mechanism with two compliances.
Clips
- Is the flexure thickness divided by twice the bend radius under three percent for your material?
- Does the flexure bend within the layer plane rather than across it?
- Does it have a geometric engagement, or does it rely purely on spring preload that will relax?
Hooks
- Is the throat depth sized from the moment, not from the visual proportion?
- Is the load line over the fixing rather than beside it?
- Is the rated load a long-term load or a short-term one? They differ by a factor of two or more.
Handles
- Is there a metal washer under every bolt head, with a flat boss behind it?
- Is the handle a closed section rather than a solid bar?
- Are the layer lines running along the pull direction rather than across it?
The thread running through all of this is that these parts are cheap to print and expensive to debug. A clamp that lets go in a workshop is an inconvenience. A hook that lets go above a workbench is a dropped tool. A handle that lets go is a fall. The geometry that prevents all three is not complicated, but it does have to be designed rather than assumed, and the three numbers worth remembering are the cube of the span, the ratio of thickness to bend radius, and the prying moment at the fixing. Get those three right and the part will grip long enough to be forgotten about, which is the highest compliment a part like this can receive.
For the parts that face outdoor sun and rain, the choice of ASA over ABS is the single decision that most affects service life. For parts that live near a machine tool or a hot engine, the choice of a high-temperature material like PPS-CF10 is what prevents a slow sag into uselessness. And for the contact faces on any of them, a printed pad in TPU 95A is the smallest change with the largest effect.