Preload based fastener torque analysis for bolted joints, in a single HTML file.
Torque Studio works out what torque to actually set on the wrench, starting from the preload the joint needs rather than from a number copied off a table. It shows the formulas before the numbers, cites a source for every input, and reports margins against everything that can go wrong: fastener yield and fracture, parent thread stripping, crushing under the head, joint slip, self loosening and separation.
Import a spreadsheet of fasteners, get a torque for every one of them, and print a paginated schedule your build team can work from.
It is one HTML file. No build, no install, no network, no dependencies, no browser storage. Open it and it runs.
Most published torque tables hide two assumptions that fail quietly.
The preload target. A very common mistake is to take a percentage of the fastener's ultimate strength and treat it as if it were a proof stress. For a material where yield sits well below ultimate, that puts the nominal preload at or above yield before anything else is considered. Preload should be set from yield, not from ultimate.
The friction condition. The nut factor K sits in the denominator of P = T / (K·D), so if the real friction is lower than assumed, the delivered preload is higher than intended. Measured nut factors for the same fastener vary by a factor of three between genuinely dry and lubricated. A table that does not state its friction condition, or that states one and uses a value belonging to another, will be wrong in one direction or the other. It is the single most influential variable in the whole calculation.
Torque Studio makes both explicit, then propagates the uncertainty through to a preload band rather than a single number.
The idea the tool is built around: there is a band of acceptable preload, not a single value, and torque has to land inside it.
Too high and you yield or fracture the fastener, strip the threads in the tapped part, or crush the material under the head. The governing bound is the lowest of those three.
Too low and the joint slips. Transverse slip is the dominant self loosening mechanism (Junker, SAE 690055): once a joint moves sideways under vibration, the thread helix ratchets and the fastener backs itself out. A low torque is its own failure mode, not a safe hedge. The required minimum comes from the applied shear, the faying friction coefficient and the number of interfaces, plus any separation requirement.
The tool derives both bounds, draws the installed band against them, and warns when the band falls outside or when no valid window exists at all.
A project holds many fasteners, organised into sub-projects. The sidebar shows the full hierarchy: project, sub-project, fastener. Drag fasteners to reorder them or to move them between sub-projects, drag a sub-project to reorder the whole block, and right click any of the three levels for rename, duplicate and delete.
Clicking a project opens the fastener table: every parameter of every fastener in one editable grid. Change a size, material or thread condition in the table and that fastener's own analysis updates with it. The table filters by text, sub-project and status, sorts by any column, and opens the full analysis on a double click. Two fields at the top of the project, the preload target as a percent of yield and the tool accuracy, apply to every fastener at once, so retargeting an entire schedule is one edit.
Real assemblies get built with whatever hardware is on hand, and the right torque changes with it. Create variants takes every unique fastener in the project and generates one sub-project per combination of thread condition and material you select, each covering the entire assembly. Build with greased A286 one day and dry stainless the next, and the schedule has a page for both.
Import reads .xlsx and .csv directly in the browser, with no library and no upload. It guesses which column is which, shows you the mapping and a live preview of the interpreted result, and lets you correct anything before committing. Section heading rows, the sort that say "Primary structure" across an otherwise empty row, become groups rather than failed parses.
Anything it cannot recognise is left at a safe default, flagged in the preview, and prefixed in the fastener name so it cannot be missed. There is a starter template in examples/fastener-list-template.csv.
Two reports, both paginated by measuring content rather than by fixed page breaks, so they never overflow no matter how much you put in.
 |
 |
Project report. Written for the person doing the assembly. Cover page with your logo, company name and classification banner, then the consolidated torque schedule in both newton metres and inch pounds with colour coded status per line, instructions for using the schedule, a findings section calling out every thin or negative margin, a compact calculation block per fastener with the wrench setting stated first, the basis of analysis with every material and friction source listed, the method and formulas, and an approval page. The preload target used is stated before the schedule table.
Fastener report. The full analysis for one configuration: summary and verdict, inputs with sources, method, calculations with values substituted, all four charts, and approval.
Both carry per page headers and footers with the classification banner, document ID and page numbering. Print to PDF from the report view. A4 and Letter are both supported.
The Company screen takes a logo, a company name, a classification banner and a footer line. These appear on the cover and in the header and footer of every page, and the cover carries a proprietary notice naming your organisation.
Prepared, Checked and Approved can each be signed. Signing records the name, an ISO timestamp, and a hash of every input. If any input changes afterwards, the hash no longer matches and the signature is marked void in both the app and the report. This is an auditable electronic record, not a cryptographic signature, and it does not replace formal approval in a quality system.
Libraries
- 56 thread sizes: UNC and UNF from #0-80 to 5/8, ISO metric coarse from M1.6 to M20, ISO metric fine from M4 to M16
- 30 fastener material conditions across A286, 300 series stainless, precipitation hardening stainless, titanium, nickel and cobalt alloys, alloy and carbon steel including ISO property classes, and non ferrous
- 26 parent materials including aluminium alloys, Super Invar and Invar, Kovar, titanium, stainless, alloy steel, magnesium, composites and polymers
- 10 thread conditions carrying measured nut factors with low and high bounds
- 9 head types and 5 washer types, which set the bearing diameter
- 9 locking and retention methods
Every selection loads its properties into a table with a source stated for each row. Every value is editable, and every input has an information note explaining what it does and where the default came from.
Analysis
- Installation torque from a target fraction of fastener yield, with a second alternative target for trade studies
- Preload band from tool accuracy, the full nut factor scatter band, and short term relaxation
- Margins against fastener yield, fastener ultimate, parent thread strip, under head bearing, joint slip and joint separation
- Prevailing torque handling, so the wrench setting is the preload torque plus whatever the locking feature absorbs
- Any candidate torque can be checked against the same analysis, for auditing a value already on a drawing
- Project wide preload target and tool accuracy, applied to every fastener with one edit
- Torque always shown in both newton metres and inch pounds, whichever system you are working in
Presentation
- Four live charts: preload stress against allowables, the preload window, sensitivity to friction, and a scale section view of the joint
- Full working shown with every value substituted, copyable as plain text
- Colour coded status in the app and in the report: green inside all limits, amber for a thin margin, red for anything needing review
- Light and dark themes, resizable panes, undo and redo, right click menus on projects, sub-projects and fasteners
 |
 |
Download index.html and open it in any modern browser. That is the whole installation.
git clone https://github.com/rohanbaba/torque-studio.git
cd torque-studio
open index.html # macOS
xdg-open index.html # Linux
start index.html # WindowsIf GitHub Pages is enabled on this repository the tool is also live at https://rohanbaba.github.io/torque-studio/.
Spreadsheet import uses DecompressionStream, so it needs Chrome 80, Edge 80, Firefox 113 or Safari 16.4 or newer. Everything else works in any current browser.
- Pick the thread series and size. Geometry loads automatically.
- Pick the fastener material and condition. Confirm the allowables against your certs, since spec minimums vary more than people expect.
- Pick the thread condition. This is the most consequential choice in the tool.
- Pick the parent material and set the thread engagement.
- Enter the applied shear and tension per fastener if the joint carries them. Without these, no minimum preload requirement can be derived and only the upper bound is checked.
- Read the total installation torque from the toolbar. It stays visible wherever you scroll, in both unit systems.
The core relation is the standard nut factor form used by NASA GSFC 540-PG-8072.1.2B:
T = K · D · P
with the target preload set as a fraction of fastener yield:
P_target = c · F_ty · A_t
GSFC uses c = 0.65. The preload actually delivered is bounded on both sides:
P_max = T (1 + a) / (K_min · D) governs yield, strip and bearing
P_min = T (1 - a) / (K_max · D) · (1 - r) governs slip, separation and self loosening
where a is tool accuracy and r is short term relaxation. Note which friction bound belongs to which failure mode. Low friction is the overload case, high friction is the slip case.
The window the installed band has to fit inside:
P_min,req = max( FS·V / (µ·n_i) , FS·T_applied )
P_max,all = min( P_y , P_strip , P_bear )
Internal thread stripping uses the FED-STD-H28 and ASME B1.1 shear area:
A_n = π · n · L_e · D_s,min · [ 1/(2n) + 0.57735 ( D_s,min - E_n,max ) ]
The bracket term is the external major diameter minus the internal pitch diameter. Substituting the external pitch diameter there is an easy mistake that inflates the area and produces a falsely positive margin.
A full derivation with every symbol defined, all data tables and their sources, is in docs/METHOD.md.
The method is checked against the worked example published in GSFC 540-PG-8072.1.2B itself, for a 10-32 A286 fastener at F_tu 200 ksi and F_ty 180 ksi:
| Quantity | Torque Studio | GSFC published |
|---|---|---|
| Target preload at 65% of yield | 2,340 lbf | 2,340 lbf |
| Torque, lubricated at K = 0.18 | 80.0 in·lb | 80 in·lb |
| Torque, dry at K = 0.30 | 133.4 in·lb | 133 in·lb |
Thread data is derived rather than transcribed, then checked against published tables. Imperial class 2A and 2B limits come from the ASME B1.1 tolerance chain and reproduce published values exactly for every size tested. Metric 6g and 6H limits are read from ISO 965-1 tables validated against eight independently confirmed anchor sizes. Every tensile stress area in the library reproduces the published value to within 0.4 percent.
Everything lives in the browser tab and nowhere else. Nothing is written to browser storage, nothing is uploaded, and closing the tab discards it.
Export writes all projects, fasteners and branding to a single JSON file. Import brings it back, or loads a fastener list from a spreadsheet. CSV exports the current schedule. Treat exported files with the same care as the drawings they describe.
Read these before using any output on hardware that matters.
- Torque values are engineering estimates for planning and review. Nut factors carry large scatter and should be substantiated by torque tension testing on the actual hardware, lubricant and tooling before flight.
- Where threaded inserts are used, the insert manufacturer's published pull out rating governs instead of the parent thread strip calculation in this tool.
- Criticality 1 and 2 joints need a full NASA-STD-5020 separation, slip and fatigue assessment in addition to this analysis.
- Not covered: fatigue, thermal preload change, fastener bending, shear and tension interaction in the fastener itself, joint stiffness and load introduction factors, creep, and preload loss over long durations.
- The separation check uses a loading plane factor of 1, which is conservative but crude. A proper joint stiffness analysis will give a less punishing answer.
- Countersunk heads bear on a cone, not on a flat annulus, so the under head bearing check does not apply to them directly and is suppressed. Check the cone bearing and countersink depth separately.
- Some parent material shear allowables are derived at 0.60 times ultimate rather than taken from a published shear allowable. Those rows say so. Confirm them before flight.
- Import guesses. Always review the preview, and never accept a row flagged as unrecognised without setting it by hand.
- Final flight torque values require sign off by the responsible structural engineer.
- NASA GSFC 540-PG-8072.1.2B, Mechanical Fastener Torque Guidelines
- NASA-STD-5020B, Requirements for Threaded Fastening Systems in Spaceflight Hardware
- NASA/TM (NTRS 20150004064), Torque Tension Testing of Fasteners used for NASA Flight Hardware Applications
- Junker, G. H., New Criteria for Self-Loosening of Fasteners Under Vibration, SAE 690055
- FED-STD-H28 and ASME B1.1, thread stress areas, stripping areas and class limits
- ISO 898-1, ISO 3506 and ISO 965-1, metric threads and property classes
- MMPDS and AMS specifications, material allowables
Issues and pull requests are welcome, particularly:
- Additional thread series, especially UNJ and metric extra fine
- Material allowables with a citable source, or corrections to existing ones with one
- Measured nut factor data for lubricants and coatings not currently listed
- Import parsers for other spreadsheet layouts
- Corrections to the analysis. If you think a formula or a number is wrong, please open an issue with the source you are working from
Keep the single file, zero dependency structure. It is the reason the tool is easy to hand to someone else.
MIT. See LICENSE.
This software is provided as is, without warranty of any kind. It is an engineering aid, not an authority. You are responsible for verifying every result before it touches hardware.