A decision twin is a digital twin built around a decision rather than a visualization: it maps live condition data onto a 3D model of the asset, forecasts how each finding evolves, prices the cost of acting now versus waiting, and routes the recommended action to the person with authority. The four examples here — a processing plant, a fighter aircraft, a crude storage tank and a bottling line — each show the same pattern: find it early, price the wait, decide inside a planned window.
How a decision twin works
↺ The twin is the interface; the decision is the product.
| Digital twin (visual) | Decision twin | |
|---|---|---|
| Answers | What does the asset look like now? | What should we do about it, by when, and who decides? |
| Output | 3D model and dashboards | Ranked options with cost, risk and authority |
| Time | Present state | Present plus 30 / 60 / 90-day forecast |
| Measured by | Model fidelity | Outages avoided, cost of waiting, decision latency |
Processing plant twin · Processing plant · Train 2
01 · Manufacturing & process
A gas-processing train — tank farm, fractionation column, compressors and utilities — with 1.2 million sensor tags feeding one condition heatmap. The decision it drives: fix the compressor bearing in a planned slowdown, before it trips the whole train.
▶ Open the live processing plant twinWhat the heatmap shows right now
| Zone | Finding | Measurement | Remaining life | Recommended action |
|---|---|---|---|---|
| K-301 · Compressor K-301 · stage 2 | Bearing-wear signature | Vibration: 7.8 mm/s (alarm 7.1) | 21 days | Change bearings in this week’s planned slowdown; pre-stage the spare kit. |
| S-2 · LPG sphere S-2 · PSV-7 | Relief-valve test overdue | Test due: 12 days ago | Overdue | Pull PSV-7 for bench test and swap in the certified spare. |
| C-101 · Column C-101 · trays 18–24 | Rising differential pressure | ΔP: +22% vs design | 60 days | Reduce reflux 4% and schedule an antifoam trial. |
| T-3 · Crude tank T-3 · floor | Floor-plate corrosion | MFL indication: 44% loss, 2 plates | 180 days | Keep in service at reduced fill height; plan floor repair at next cleaning. |
| PR-3 · Pipe rack PR-3 · 8" crude line | Corrosion under insulation | Wall loss: 31% at 3 locations | 140 days | Strip insulation at the three hot spots; confirm with UT. |
The cost of waiting
Fast-forward the twin and watch the same finding become a different decision. Illustrative figures.
| Horizon | Readiness | Risk exposure | Act now | Wait to failure |
|---|---|---|---|---|
| Today | Running · 2 alerts | $3.8M | $180K · 36h | $4.2M · 6d |
| +30 days | Running · throughput −4% | $5.2M | $240K · 36h | $4.2M · 6d |
| +60 days | At risk · trip likely | $8.9M | $1.1M · 3d | $4.2M · 6d |
| +90 days | Tripped · Train 2 down | $14.6M | — | $4.2M · 6d |
The decision it puts in front of a person
- a. Run to the planned turnaround — Keep Train 2 at full rate until March. K-301 vibration keeps climbing.
- b. Swap K-301 bearings in this week’s planned slowdown — 36-hour slowdown, parts on site. Reliability and operations approve. ← recommended
- c. Controlled shutdown · full compressor overhaul — Six days down, contractor mobilization, $4.2M of deferred production.
Data it reads: Historian (PI, Aveva, Ignition), CMMS/EAM (SAP PM, Maximo), production plan, stores and spares, inspection reports.
Who decides: Reliability engineer and operations manager.
Fighter aircraft twin · Multirole fighter · Tail 412
02 · Defense & aerospace
One airframe of a multirole fleet, with structural, propulsion, landing-gear and flight-control findings on the model where maintainers actually work. The decision it drives: restrict and inspect now, or ground later at ten times the cost.
▶ Open the live fighter aircraft twinWhat the heatmap shows right now
| Zone | Finding | Measurement | Remaining life | Recommended action |
|---|---|---|---|---|
| WR-01 · Wing root fitting · starboard | Fatigue crack indication | Crack length: 1.9 of 3.0 mm limit | 38 flight hours | Restrict to 5G and inspect within 12 flight hours. |
| NZ-03 · Nozzle petals · hot section | Thermal distress | Hot-section temp: +46 °C vs baseline | 120 flight hours | Borescope at the next phase inspection; trend exhaust temperature every flight. |
| GL-04 · Main landing gear · port | Hydraulic seep at seal | Pressure decay: 3.1 psi/hr | 64 flight hours | Replace the strut seal kit at the next turnaround. |
| IN-02 · Intake lip · lower | FOD erosion | Erosion depth: 0.42 mm | 210 flight hours | Blend and recoat during the next scheduled maintenance. |
| VT-06 · Vertical tail attach | Corrosion pitting | Pitting depth: 0.18 mm | 300 flight hours | Treat and seal; re-inspect in 90 days. |
The cost of waiting
Fast-forward the twin and watch the same finding become a different decision. Illustrative figures.
| Horizon | Readiness | Risk exposure | Act now | Wait to failure |
|---|---|---|---|---|
| Today | Mission capable · restricted | $2.4M | $48K · 2d | $1.9M · 21d |
| +30 flight hours | Mission capable · restricted | $3.1M | $62K · 2d | $1.9M · 21d |
| +60 flight hours | Restricted · 3G envelope | $4.6M | $410K · 9d | $1.9M · 21d |
| +90 flight hours | Not mission capable | $7.8M | — | $1.9M · 21d |
The decision it puts in front of a person
- a. Fly as scheduled — Keep the full envelope. The fatigue crack keeps growing at the wing root.
- b. Restrict to 5G · inspect wing root within 12 FH — Stays mission capable. Maintainer and engineering approve. ← recommended
- c. Ground · replace wing fitting now — 21 days down, spares pulled from depot.
Data it reads: Maintenance records, flight-hour logs, NDI and borescope results, health and usage monitoring, parts and depot availability.
Who decides: Maintenance officer, engineering and operations.
Storage tank + crawler twin · Crude storage tank T-2
03 · Oil & gas · robotics
An API 650 crude tank mapped by a wall-climbing crawler — 214,880 ultrasonic readings, plus drone and survey data — so shell loss at a weld becomes a decision about fill height, not a surprise leak.
▶ Open the live storage tank + crawler twinWhat the heatmap shows right now
| Zone | Finding | Measurement | Remaining life | Recommended action |
|---|---|---|---|---|
| W-07 · Shell course 2 · weld 7 | Wall loss at vertical weld | Wall thickness: 5.8 of 10 mm | 45 days | Hold fill at 70% and confirm with manual UT before Friday’s fill cycle. |
| AN-02 · Annular plate · south | Edge settlement | Differential settlement: 38 mm | 150 days | Survey again in 60 days; seal the chime to stop water ingress. |
| MOV-1 · Outlet valve MOV-1 | Seat leak-through | Leak rate: 0.8 L/min | 70 days | Replace seat at the next line outage; lock the upstream valve until then. |
| PV-3 · Pressure-vacuum vent PV-3 | Pallet sticking | Set-point drift: +14% | 90 days | Clean and re-test PV-3; add it to the monthly drone route. |
| RF-04 · Roof plates · north-west | Pitting under coating | Pit depth: 1.2 mm | 240 days | Spot-repair coating in the spring campaign. |
The cost of waiting
Fast-forward the twin and watch the same finding become a different decision. Illustrative figures.
| Horizon | Readiness | Risk exposure | Act now | Wait to failure |
|---|---|---|---|---|
| Today | In service · fill limited | $1.1M | $36K · 0d | $1.2M · 28d |
| +30 days | In service · fill limited | $1.4M | $58K · 0d | $1.2M · 28d |
| +60 days | In service · 50% fill | $2.3M | $420K · 12d | $1.2M · 28d |
| +90 days | Out of service · leak risk | $3.9M | — | $1.2M · 28d |
The decision it puts in front of a person
- a. Keep filling to full height — No change to operations. Shell loss at weld 7 keeps advancing.
- b. Hold fill at 70% · confirm with UT before Friday — Crawler already mapped it. The integrity engineer approves. ← recommended
- c. Drain, clean and repair now — 28 days out of service; 118,000 bbl of storage lost.
Data it reads: Robotic UT and drone inspection data, tank records, fill schedules, survey data, API 653 limits.
Who decides: Integrity engineer and terminal operations.
Bottling line twin · Bottling line 3 · 36,000 bph
04 · Food & beverage
A 36,000-bottles-per-hour line — blow molding, filling, inspection, labeling, packing and palletizing — watching 4,200 line signals. The decision it drives: a 25-minute valve fix in today’s clean-in-place window instead of a stopped line and product on hold.
▶ Open the live bottling line twinWhat the heatmap shows right now
| Zone | Finding | Measurement | Remaining life | Recommended action |
|---|---|---|---|---|
| FV-14 · Filler · valve 14 | Fill-level drift | Under-fill: 1.9% of bottles (limit 0.5%) | 9 days | Swap valve 14 seals during today’s 2 pm CIP window; hold nothing, reject stays automatic. |
| BM-2 · Blow molder · oven zone 4 | Lamp degradation | Wall-thickness variance: +18% vs baseline | 30 days | Replace two IR lamps at the weekend changeover; raise zone 4 power 3% until then. |
| VI-1 · Empty-bottle inspector | Reject rate climbing | Rejects: 0.9% (normal 0.3%) | 21 days | Link rejects to blow-molder zone 4 — same root cause, one fix. |
| LB-3 · Labeler · glue station | Glue temperature swing | Temp swing: ±9 °C (spec ±3) | 45 days | Replace the glue-tank thermocouple; recheck label adhesion at shift start. |
| CIP-1 · Water treatment · RO skid | Conductivity creeping | Permeate: 31 µS/cm (limit 40) | 60 days | Schedule membrane clean-in-place next week; product water stays in spec. |
The cost of waiting
Fast-forward the twin and watch the same finding become a different decision. Illustrative figures.
| Horizon | Readiness | Risk exposure | Act now | Wait to failure |
|---|---|---|---|---|
| Today | Running · 2 alerts | $310K | $4K · 25 min | $96K · 10h |
| +7 days | Running · OEE −3 pts | $420K | $6K · 25 min | $96K · 10h |
| +14 days | At risk · compliance | $690K | $38K · 4h | $96K · 10h |
| +30 days | Stopped · lots on hold | $1.2M | — | $96K · 10h |
The decision it puts in front of a person
- a. Run to the weekend changeover — Keep line 3 at 36,000 bph. Under-fills on valve 14 keep climbing.
- b. Swap valve 14 seals in today’s 2 pm CIP window — 25 minutes inside a planned stop. Maintenance and quality approve. ← recommended
- c. Stop the line for a full filler rebuild — Ten hours down, contractor on site, a day of lost production.
Data it reads: PLC and line MES tags, fill-level and vision inspection, CIP schedule, quality holds, maintenance history.
Who decides: Line manager, maintenance and quality.
The pattern across all four
Find it while it’s cheap
Each twin surfaces the finding weeks before failure — a bearing signature, a crack indication, wall loss at a weld, a drifting fill valve.
Price the wait
The fast-forward turns a finding into money and readiness, so “later” stops looking free.
Use the planned window
Every recommendation fits a window that already exists: a slowdown, a turnaround, a CIP cycle, a phase inspection.
Keep the owner in charge
The twin recommends; the reliability engineer, maintainer or line manager decides — and the outcome trains the next call.
- A twin that doesn’t change a decision is a screensaver.
- Forecast the cost of waiting — that’s what moves people.
- Route every recommendation to a named owner with authority.