Plastic part design calculator: tooling axis, draft, ribs, screw bosses and dog houses

The draft angle of a face is 90° minus the angle between its normal and the tooling axis, so a wall parallel to the pull has 0° and a wall leaning back is an undercut. In a section, the bisector of the two extreme walls is the axis that maximises the smallest draft: 90° minus half the angle between their normals.

Draw a Class-A section or list the faces: the calculator finds that axis, colours each face, shows how far a "by eye" Z axis is off, and checks the 3D case, where two section bisectors are not always the best axis. A second tab sizes ribs, screw bosses and dog houses, with each guide's numbers side by side. Skill-Lync read 7,722 public plastic trim reports from its Automotive Plastic Design using CATIA V5 course: 59% use the bisecting method. Free, runs in your browser.

DuPont General Design Principles · Stratasys Direct (2017) · Protolabs · CATIA V5 draft analysis · checked against a separate solution (1,375 checks at build) · Updated 22 Nov 2026

Outer walls drafted 1.5° on the left and 7.5° on the right for a pull along Z. Z by eye fails the left wall; the bisector of the two outer walls tilts the axis 3° and gives both 4.5°.

Best axis−3.00°from Z, bisector of the extreme walls
Smallest draft there4.50°90° − 171.0°/2
At your pull1.50°minimum 3°
1.5°87.2°8.0°8.0°87.0°7.5°best −3.0°pull 0.0°Section coloured by draft · dashed: extreme walls, bisector-20..-3°-3..0°0..3°3..20°|θ| ≥ 20°
-20°-10°0°-35°-20°-5°10°25°minimum 3°best 4.50°Z 1.50°smallest draft vs pull angle · shaded: clears the minimum
  • Your pull at 0.00°: smallest draft 1.50° on segment 2 (minimum 3°).
  • Best axis: −3.00° from Z, the bisector of segments 2 and 8 (their normals span 171.00°). Smallest draft there: 90° − 171.00°/2 = 4.50°.
  • Any pull from −4.50° to −1.50° clears 3°.
  • Z by eye: smallest draft 1.50° (segment 2), 3.00° less than the best axis.

How do you find the tooling axis for a plastic part?

Cut the Class-A surface with a plane, take the two walls on opposite sides that lean closest to the pull, extend them until they meet and bisect the angle between them. That bisector is the tooling axis for the section. It is not a rule of thumb: in a section it is exactly the axis that gives every face the most draft.

The draft angle of a face is measured from the pull direction d to the face itself. With n the face's outward unit normal (out of the A side, towards the cavity):

A flat top facing the pull reads 90°, a wall parallel to it 0°, a face leaning back is negative: an undercut. In a section every segment's normal sits at an angle β from Z, and the draft at a pull angle φ is 90° − |β − φ|. The smallest draft is set by the two normals farthest from φ, so the best φ sits halfway between the two extreme normals:

S is the angle between the two extreme normals. The normals of two walls are perpendicular to the walls, so halfway between the normals is halfway between the walls: the course's bisector. For the switch bezel above, the outer walls are drafted 1.5° and 7.5° for a pull along Z: Z gives 1.50°, below 3°. The extreme normals are 171.00° apart, so the bisector sits 3.00° off Z and gives both walls 4.50°.

Any axis inside a window around the bisector still clears the minimum, and the window closes when the spread reaches 180° − 2θmin, which is 174° for 3°:

The coin holder pocket has two 2° walls facing each other: their normals span 176°, so no axis gives 3° and the walls themselves need more draft. The door trim with a wrap-over edge is worse: its wrap faces back, the best axis still leaves −6.5°, and that face needs a lifter or slider, or the split line moves to the wrap's edge. The base bracket shows the other common case: one wall exactly vertical, which Z leaves at 0°; the bisector tilts 5° and gives 5° to both.

Is the bisector in two sections the best 3D tooling axis?

Not always. Many reports bisect in an XZ section, then in a YZ section, and combine the two tilts. That only sees the faces those two sections cut. In 3D the best axis is the centre of the smallest cone (a cap on the unit sphere) that holds every face normal, and its smallest draft is 90° minus the cap's radius:

The calculator finds it exactly: the optimum is equally far from one, two or three "active" normals, so it tests every such direction. The two-section axis it compares against is the line through both tilts:

On the bezel with four 6° walls and a corner chamfer at −1°, sections through the middle never cut the chamfer: the two bisectors stay on Z and the chamfer stays at −1.00°. Tilting −2.90° in XZ and −2.90° in YZ, towards the corner, gives every face 3.10°. On the box with walls at 2°, 4°, 6° and 8°, the two sections give 3.998° and the true optimum 4.000°: close, because the faces line up with the sections.

What do the colours in a CATIA draft analysis mean?

Whatever the colour scale says: the colours are bands between the values on the scale, and you can type your own. Read the scale before you read the part. 11 reports quote a six-band table (green 0 to 3°, yellow 0 to −3°, cyan 3 to 20°, blue −3 to −20°, red beyond ±20°); with it, green means less than 3°, and faces that pass are cyan, or red above 20° (flat tops). 1,459 reports mention green next to 3°, mostly after re-setting the scale so that green starts at 3°; then green is a pass.

BandSix-band tableGreen from the minimum up
20° to 90° redgreen (passes 3°)
3° to 20° cyangreen (passes 3°)
0° to 3° greenamber (under 3°)
−3° to 0° yellowred (undercut)
−20° to −3° bluered (undercut)
−90° to −20° redred (undercut)

We could not read Dassault Systèmes' own V5 page on the draft analysis this time (the catiadoc.free.fr mirror refuses requests and is not archived), so we cannot say whether the six-band table is CATIA's Full-mode default. A CATIA V5R14 training sheet (CEDIS / DENC) sets Quick mode to green between −0.99° and 0.99° and red or blue beyond ±1°, with the pull along the compass Z by default. Skill-Lync's draft analysis guide walks through the dialog.

How much draft does a plastic trim part need?

3° on the Class-A side is the course's rule, and 4,121 of the 7,722 reports quote it. It is a trim-part rule, not a law of moulding: for smooth faces DuPont's design guide (Table 3.01) gives 0 to ¼° on shallow PBT and acetal and ½° to 1° on deep reinforced nylon, and Protolabs strongly advises at least 0.5° with 1° to 2° working in most cases. What pushes trim to 3° and beyond is grain: texture is a field of small undercuts. Protolabs asks for 3° on a light texture (PM-T1) and 5° or more on a heavy one (PM-T2); depth-based rules give:

Texture depthMold-Tech (2020): 1° per 12.5–15 µmDuPont / Machinery's Handbook: 0.5° + 1° per 0.025 mmOlder rule of thumb: 1.5° per 0.025 mm
0.025 mm1.67–2.00°1.50°1.50°
0.05 mm3.33–4.00°2.50°3.00°
0.1 mm6.67–8.00°4.50°6.00°

The texture rules are the ones in the injection moulding calculator (draft tab), which also covers clamp force, cooling time and shrinkage. 112 reports call 3° an industry standard or rule of thumb; with a 0.05 mm grain the rules above ask for 2.5° to 4°.

How thick should a rib be, and what is rib root thickness?

Thinner than the wall, so the junction does not become a thick spot that cools last and pulls a sink mark into the A side. The guides we read agree on the idea and differ on the number:

SourceRib baseOther rib rules
DuPont, General Design Principles, ch. 4½ to ⅔ of the wall¼° to ½° taper on deep ribs; rib = wall with r = 0.5 t gives a 1.5 t circle
Stratasys Direct (2017)at most 60% (50 to 60% in another section)height under 3 × rib thickness; draft at least 0.25° to 0.5° per side
Protolabs, Injection Molding Basics40 to 60%no height or draft figure
Course rule, as 598 reports quote it40% of the wallheight up to 5 t, 0.5° per side, tip at least 0.75 mm, fillet up to 0.25 mm, root circle within ±30% of t

Draft thins the rib towards its tip, and the course's 0.75 mm tip minimum then caps the height. The course rib, 1 mm on a 2.5 mm wall, 12.5 mm tall at 0.5° per side:

Above 14.3 mm that rib drops under 0.75 mm. "Rib root thickness" is the diameter of the largest circle in the junction, which CATIA users draw as a tritangent circle on a section. With straight rib sides and root fillets r, its centre sits on the rib axis where the distance to the far face equals the distance to the fillets:

For the course rib with 0.25 mm fillets that is 2.6875 mm, 1.075 t; 64 attachment-feature reports give a measured value of about 2.686 mm, where the 0.5° draft on the rib sides takes off the last µm. A rib as thick as the wall with r = 0.5 t gives exactly 1.5 t, DuPont's figure. 33 reports copy the rule as "not more than 30% of the wall thickness" (1, 2, 3): read literally that asks for a circle under 0.75 mm, smaller than the wall itself, which no junction can have. The rule means within 30% of the wall, at most 1.3 t = 3.25 mm, as 64 reports write it.

What size should a screw boss be?

For self-tapping screws, DuPont's design guide (ch. 9) gives a hole equal to the screw's pitch diameter for the best ratio of stripping to driving torque, a boss about 2.5 times the screw, and notes that stripping torque stops rising at about 2.5 pitch diameters of engagement (its pull-out tables use 2.5 × the screw diameter):

Stratasys Direct sets the sink side: boss wall under 60% of the nominal wall, base radius at least 0.25 t. The two rules cannot both hold on a thin wall. An M3 boss by DuPont's numbers has a 2.65 mm wall at the base, 106% of a 2.5 mm wall: fine off the show surface, a sink risk on it, which is why bosses are cored, tied to walls with ribs and kept away from the A side. Course reports size the boss from the wall instead: 302 quote a boss wall of 40% of the nominal wall and 163 60%, which for an M6 screw in a 5.2 mm hole gives a 7.2 to 8.2 mm boss, about half of DuPont's 2.5 × 6 = 15 mm. We could not read EJOT's PT screw design data (its site refused our requests), so no per-material screw table is given here.

What is a dog house in plastic part design?

A dog house is a small open box moulded on the B side to carry a clip or a push pin, so the clip load goes into the part without a boss. None of the guides we read gives dog-house numbers, so the calculator treats its walls like ribs. The course rules, as reports quote them: the wall where the dog house meets the part (c) at most 40% of t (440 reports), a minimum width of 12 mm (103) and a height of 3 to 6 mm for the clip seat (291). At 2.5 mm that is a 1 mm wall; with 0.5° per side over 8 mm it ends at 0.860 mm.

What did 7,722 student reports write?

Skill-Lync matched 9,693 public reports titled base bracket, switch bezel, coin holder, bumper, door trim, door applique, arm rest, fan cover draft analysis or tooling axis on 22 Nov 2026; 11 crash and mesh bumpers and 57 Alias concept-modelling bumpers were left out, leaving 9,625. 7,722 have text; 1,903 are image-only. The method each text uses for the tooling axis:

Tooling axis method in the textReportsExamples
Bisecting two walls (lines or sketches)4,5691, 2, 3
A line normal to a plane or flat face3071, 2, 3
Z kept, or the compass moved until most faces turn green1991, 2, 3
A tooling axis, method not stated2,5371, 2, 3
No tooling axis in the text1101, 2, 3
Total with text7,722
  • Minimum draft. 4,121 quote 3°; 112 call it a standard value or an industry rule (1, 2, 3). 11 quote the six-band colour table; 1,459 mention green next to 3°.
  • Wall. Taking the first thickness in mm each text gives, 2,739 reports use 2.5 mm and 730 use 3 mm; 4,093 give none in the text.
  • Attachment features. Of 3,491 Week 9 and 10 attachment-feature reports, 2,915 have text: 598 quote ribs at 40% of the wall, 387 a height up to 5 t, 161 the 0.75 mm tip, 470 check the rib root thickness and 34 write out a tip of about 0.78 mm.

Counts as of 22 Nov 2026, n = 7,722 trim reports and 2,915 attachment-feature reports with text, from the text of each public report: Skill-Lync's own data, a starting point, not a grade. Numbers inside images are not counted, and a report is counted once per row.

  • Bisecting method: in one section the bisector of the two extreme walls is exactly the axis that maximises the smallest draft. If the part has walls on all four sides, check a second section (or the 3D tab): two section bisectors combined are not always the 3D optimum.
  • Colour scale re-set so green starts at your minimum: then green is a pass. State the limits you typed in, so a reader can tell which table you used.
  • Rib base 1 mm is 40% of the 2.5 mm wall, within your stated 40%.
  • 40% of the wall is outside ½ to ⅔ of the wall (DuPont).
  • Tip after 0.5° per side over 12.5 mm: 0.782 mm (78% of the base), against the course's 0.75 mm minimum at the tip. We did not find a tip minimum in the guides we read.
  • "Rib root thickness not more than 30% of the wall" cannot be met as written: the circle at a rib root is never smaller than the wall (2.5 mm), so 2.686 mm is not under 30% (0.75 mm). The rule means within 30% of the wall: at most 1.3 t = 3.250 mm (33 public reports did this; examples 1, 2, 3).
  • For a 1 mm rib with 0.25 mm root fillets on a 2.5 mm wall the inscribed circle works out to 2.688 mm (straight sides; the 0.5° draft takes off a few µm), which matches your measurement.
  • Height 12.5 mm is more than 3 × the rib thickness (3.0 mm), Stratasys Direct's limit; against the wall it is 5.0 t.
  • Screw boss for a 6 mm screw: hole 5.2 mm = 0.87 Ds (DuPont: about 0.8 Ds), OD 7.2 mm = 1.20 Ds (DuPont: 2.5 Ds).

What this calculator does not cover

  • Real surfaces. A section is a polyline and a 3D part a list of face normals. CATIA's draft analysis checks every point of the surface; use this to choose and defend the axis, then run the analysis.
  • Sliders, lifters and split lines. When no axis clears a face, the calculator says so; designing the side action is the next step.
  • Sink and warpage. The root circle ranks junctions; how deep a sink gets depends on the resin, packing and gate, which a fill simulation predicts.
  • Heat stakes and per-material screw tables. We did not find and read a heat-staking supplier's design numbers or EJOT's PT tables, so neither is here.
Learn plastic trim design

The tooling axis, draft analysis, ribs, screw bosses and dog houses are the Week 8 to 10 challenges of Skill-Lync's plastic design course, on real trim parts such as a switch bezel and a door trim lower.

Frequently asked questions

How do you find the tooling axis for a plastic part?

Cut the Class-A surface with a plane, find the two walls on opposite sides that lean closest to the pull direction, extend them and bisect the angle between them. In that section the bisector is the direction that gives every face the most draft: 90° minus half the angle between the two walls' normals.

What is the minimum draft angle for a plastic trim part?

The Skill-Lync CATIA plastic course asks for 3° on the Class-A side, which covers grained trim. Guides for smooth faces allow less: Protolabs strongly advises at least 0.5° and says 1° to 2° works in most cases. Texture adds about 1° for every 0.025 mm of depth, or more by Mold-Tech's rule.

Why can no tooling axis give 3° of draft on my part?

Because two walls face each other too closely. If the normals of the two extreme walls are more than 174° apart, at least one of them gets less than 3° whatever the axis. Add draft to those walls, move the split line, or free that face with a slider or lifter in the tool.

How thick should a rib be compared with the wall?

Guides give 40% to 66%: DuPont ½ to ⅔ of the wall, Stratasys Direct at most 60%, Protolabs 40 to 60%. The course rule is 40%, 5 t tall, 0.5° per side and at least 0.75 mm at the tip. A thinner rib keeps the junction circle small, which is what keeps the sink mark off the A side.

What is rib root thickness and how is it checked?

It is the diameter of the largest circle that fits in the rib-to-wall junction, drawn in CATIA as a tritangent circle on a section. For a 1 mm rib with 0.25 mm fillets on a 2.5 mm wall it is 2.688 mm, 1.075 times the wall. Keep it within 1.3 times the wall.

What size should a screw boss be for a self-tapping screw?

DuPont's design guide gives a hole about equal to the screw's pitch diameter (0.8 × screw diameter), an outside diameter of 2.5 × the screw and about 2.5 × the diameter of thread engagement. Stratasys Direct asks for a boss wall under 60% of the part wall, so bosses are cored and tied to walls with ribs.

Related on Skill-Lync

Sources

  • DuPont, General Design Principles for DuPont Engineering Polymers, Module I: ch. 3 Table 3.01 draft angles and "add 1° per 0.025 mm" for texture, boss OD 2 to 3 × the hole; ch. 4 ribs ½ to ⅔ of the wall, ¼° to ½° taper, the 1.5 T inscribed circle (Fig. 4.07); ch. 9 self-tapping screws: hole ≈ pitch diameter (0.8 Ds), boss 2.5 × the screw, torque levelling off at about 2.5 pitch diameters, tables at L = 2.5 Ds. Read 22 Nov 2026.
  • Stratasys Direct, Injection Molding Design Guidelines (2017), pp. 3-5: rib at most 60% (50 to 60%) of the wall, rib height under 3 × its thickness, rib draft at least 0.25° to 0.5° per side, boss wall under 60%, base radius at least 0.25 t. Read 22 Nov 2026.
  • Protolabs, Injection Molding Basics: rib-to-wall ratio 40 to 60%; Draft: 0.5° on vertical faces strongly advised, 1° to 2° in most cases, 3° for light texture (PM-T1) and shut-offs, 5° or more for heavy texture (PM-T2). Read 22 Nov 2026.
  • Standex Engraving Mold-Tech technical data sheet (2020) and the older 1.5° per 0.025 mm rule, as used by the injection moulding calculator's texture table (src/lib/moulding.ts).
  • CEDIS GmbH / DENC AG, CATIA V5R14 training sheet 2-47 "Schrägenanalyse - Draft Analysis" (cad.de): Quick and Full mode, the ±1° Quick-mode example, the compass Z as the default direction. Not read: Dassault Systèmes' CATIA V5 user guide page on the draft analysis (catiadoc.free.fr returned 403 and has no archived copy); IBM APARs HD62555 and HD77620 (draft analysis colour display, 0.1° precision) were read and give no default ranges.
  • Not read: EJOT PT and DELTA PT design flyers and Covestro's Part and Mold Design guide (both sites refused our requests); no numbers from them are used.
  • Skill-Lync, 9,693 public plastic trim reports and 3,491 attachment-feature reports, read 22 Nov 2026. scripts/verify/plastic_part.py recomputes every case another way (draft from the angle between normal and pull by atan2, the section optimum by a 0.05° scan and a ternary search, the bisector from line intersections, the 3D optimum as the minimum-norm point of the normals' convex hull with a sphere-grid cross-check, the root circle by a 2D search on the section's lines and arcs, the report rules re-implemented from the text), 186 cases; the page checks itself against it at build time (1,375 checks).
Free counselling

From a draft analysis to a plastic design job.

Being able to say why a tooling axis is the right one, and what a rib does to the A side, is what trim and interior teams ask about. A Skill-Lync counsellor can map the course that fits where you are now.

Talk to a plastic design course counsellor

Free. We call you back, usually the same day.

By submitting you agree to our Terms and Privacy Policy.

Talk to a career counsellor

Talk to a career counsellor

Free 20-minute call. We'll map the right course to where you want to work.

By submitting you agree to our Terms and Privacy Policy.