CT Scan Reverse Engineering
You have a hydraulic manifold with undocumented intersecting galleries, a legacy pump housing with no surviving drawings, or a turbine component with sub-millimeter internal cooling passages. A structured-light system will give you a precise point cloud of the outside and nothing else. This article walks through exactly how industrial CT scanning fills that gap - the physics, the workflow, the accuracy limits, the cost math, and the decision logic for determining whether CT is the right call before ever quoting a project.
For context on where CT fits within our broader practice, start with the foundations of reverse engineering primer, then see our full reverse engineering services overview. This article goes deep on the CT-specific detail.
Why Surface Scanning Alone Fails on Internal Geometry
Every structured-light scanner, every laser tracker, every handheld scanner - including high-end units common across the industry such as phase-based terrestrial scanners from Leica, FARO, and Trimble - operates on line-of-sight physics. The sensor emits light or a laser beam, that beam hits a surface and reflects back. If it cannot reach the surface, the surface does not exist in your dataset.
That means any undercut, blind bore, internal coolant channel, O-ring groove at the bottom of a deep pocket, or sealed void is simply absent from the scan. The point cloud looks complete on screen. It is not.
The traditional workaround is destructive sectioning. Wire EDM or a precision saw slices the part into cross-sections. Each slice is scanned individually, then the slices are registered back together digitally. The cost structure of this approach is punishing:
- Wire EDM or precision saw time: $150-400 per cut
- Each cross-section scan session: $200-600 depending on geometry
- Assembly and registration labor: $75-150/hr × 4-12 hours = $300-1,800
- Accumulated registration error across reassembled slices: ±0.1-0.5 mm
- And you have destroyed the only sample. If that part is a legacy casting, a discontinued OEM component, or the last surviving prototype, the replacement cost is effectively infinite.
| Capability | Laser/Structured-Light Scan | Industrial CT Scan |
|---|---|---|
| External surfaces | Full capture, ±0.02-0.05 mm | Full capture, ±0.05-0.2 mm |
| Wall thickness measurement | Not possible (single surface only) | Direct measurement, both walls captured |
| Undercuts | Partial or none depending on angle | Full capture regardless of orientation |
| Enclosed voids / internal channels | Not captured | Fully captured |
| Blind bores and threaded forms | Partial (open end only) | Full thread form, depth, and root geometry |
| Press-fit interfaces and internal shoulders | Not captured | Captured with ±0.05-0.2 mm accuracy |
| Multi-material assemblies (no disassembly) | Not possible | Density-contrast segmentation isolates each material |
How Industrial CT Scanning Works for Reverse Engineering
Industrial CT scanning borrows the same foundational physics as medical CT - X-rays pass through the object, and the differential absorption of those rays through different materials produces contrast - but at much higher energies and with a fixed rotating geometry designed for dimensional metrology rather than soft-tissue imaging.
The mechanical setup: the part sits on a precision rotation table between an X-ray source and a flat-panel detector array. The source and detector are fixed; the table rotates. At each angular increment, the detector captures a 2-D projection image of the X-ray shadow passing through the part. Collect 1,000-3,600 of those projections across a full 360-degree rotation, feed them into a reconstruction algorithm, and the output is a 3-D volumetric voxel dataset - every cubic unit of the part assigned a density value.
Reconstruction algorithm selection matters in practice. The Feldkamp-Davis-Kress (FDK) algorithm - the cone-beam variant of filtered back-projection - is the standard choice for most industrial CT work: fast, predictable, and adequate when signal-to-noise ratio is high (thin-walled aluminum, polymer housings, small castings). Iterative reconstruction - typically SART (Simultaneous Algebraic Reconstruction Technique) or CGLS (Conjugate Gradient Least Squares) - is the right choice when pushing the system’s penetration limit: thick steel sections, high-density alloys, or multi-material assemblies where streak artifacts from FDK would degrade segmentation. Iterative methods require 4-8× more compute time but recover usable geometry from projections that FDK cannot cleanly reconstruct.
Key system parameters that determine what CT can do on your part:
- Voltage (kV) and power: 100-225 kV for small aluminum and polymer parts; 225-450 kV for medium steel assemblies up to roughly 80 mm wall thickness; 1-9 MeV linear accelerator (linac) systems for large steel castings up to 300-400 mm section thickness. Systems like the Nikon XT H 450 (450 kV, 450 W) cover the medium-to-heavy industrial range and handle the majority of hydraulic manifold and cast housing work. The ZEISS METROTOM 1500 is a high-accuracy CT metrology platform optimized for traceable dimensional measurement at 225 kV, well-suited to turbine components and medical devices where traceable accuracy documentation is required; ZEISS specifies its MPE(SD) as 4.5 + L/50 µm per VDI/VDE 2630 Sheet 1.3. For very large assemblies, high-energy industrial CT systems with meter-scale fields of view are available when the manifold or casting exceeds what a standard industrial tube can envelop.
- Voxel resolution: Micro-CT systems (parts under 50 mm) achieve 1-5 µm voxels. Standard industrial CT on medium assemblies (up to ~600 mm diameter) runs 50-200 µm voxels. The voxel size directly governs the finest feature you can resolve - a 0.5 mm coolant passage requires voxels no coarser than 50 µm to characterize wall geometry with confidence.
- Detector panel size and rotation table payload: Larger detectors allow bigger parts in a single scan position; payload ratings typically 5-150 kg on industrial systems.
Output formats: The raw output is a DICOM or VGI volume stack - essentially thousands of 2-D slice images stacked into a 3-D volume. Software packages including VGStudio MAX, Volume Graphics Standard, and Dragonfly convert this volume to a usable mesh or point cloud via thresholding and segmentation. The most powerful feature at this stage: density-based segmentation can isolate individual material phases. An aluminum housing, its steel inserts, and its rubber seals each absorb X-rays at different rates. VGStudio MAX reads those contrast differences and lets each component be labeled and extracted independently - from a single scan, without touching a wrench.
Accuracy Benchmarks: What CT Can and Cannot Achieve
CT reverse engineering accuracy is governed by ISO 10360-11, which extends the coordinate metrology framework to CT systems. For a calibrated industrial system running on steel or aluminum, traceable volumetric accuracy is typically ±0.05-0.2 mm. That bracket matters: 0.05 mm is achievable on a thin-walled aluminum casting at an optimal kV setting; 0.2 mm is realistic for a dense steel assembly near the system’s penetration limit.
Micro-CT on polymer and composite parts brings that down further: ±0.01-0.03 mm is achievable for parts under 50 mm - which is why fuel injector nozzle geometry, medical device catheter lumens, and porous implant lattice structures are routine CT reverse engineering candidates.
The accuracy killers to know before you commit to CT:
Beam hardening: X-rays passing through dense material are preferentially absorbed at lower energies, which causes the reconstruction to overestimate density at the edges and underestimate at the center - visible as a “cupping” artifact on the density histogram. VGStudio MAX offers two correction paths: polynomial BHC, which applies a generalized polynomial correction curve to the projection data and works well for single-material homogeneous parts like aluminum housings or steel shafts; and empirical BHC, which derives the correction from an actual scan of a reference calibration body and is the right call for multi-material assemblies or parts where the polynomial assumptions break down. For titanium or hardened steel sections thicker than 50 mm, even with empirical BHC applied, plan for ±0.1-0.3 mm in those dense zones - the artifact is reduced, not eliminated.
Scatter and ring artifacts: Scatter from high-density inclusions creates halo patterns around dense features; detector element defects produce ring artifacts in the reconstruction. These must be flagged by the technician before CAD modeling begins - a trained operator will identify them in the volume review and note which features are artifact-influenced in the project report.
| Method | Accuracy Range | Internal Feature Access | Notes |
|---|---|---|---|
| Structured-light / laser scan | ±0.02-0.05 mm | Line-of-sight only | Best surface accuracy, no internal access |
| Industrial CT scan | ±0.05-0.2 mm | Full volumetric access | Sweet spot for internal geometry |
| CMM (touch-probe) | ±0.003-0.01 mm | Probe-accessible features only | Highest accuracy, cannot reach enclosed features |
| Micro-CT (small parts <50 mm) | ±0.01-0.03 mm | Full volumetric access | Near-CMM accuracy for small parts |
Practical rule: if any GD&T tolerance on an internal feature is tighter than ±0.05 mm, validate CT output with a calibrated reference artifact or follow up with CMM probing on any features the CMM can physically reach. CT is the geometry capture tool; CMM is the high-stakes validator.
The CT-to-CAD Workflow Step by Step
Understanding how the scan-to-CAD workflow builds parametric solids from scan data in the context of surface scanning is useful background - the CT version adds pre-scan assessment and volume reconstruction steps before reaching the mesh-to-CAD phase that surface scanning enters directly.
Step 1 - Pre-scan assessment: Material, wall thickness, density map, maximum cross-section. This determines kV setting, estimated scan time (15 minutes for a small polymer housing, up to 4 hours for a large steel casting), and whether multi-orientation scanning is needed to reduce cone-beam artifacts at the top and bottom of the part.
Step 2 - Volume acquisition: Part fixtured on the rotation table with anti-vibration padding if needed. 1,000-3,600 X-ray projections captured at incremental angles. Scan time depends on required signal-to-noise ratio and part size.
Step 3 - Volume reconstruction: Raw projections processed into a 3-D voxel volume using VGStudio MAX or Dragonfly. FDK reconstruction for standard geometry; SART or CGLS iterative reconstruction for high-density or difficult multi-material parts. Beam hardening correction (polynomial or empirical), scatter correction, and ring artifact suppression applied during this step. Output: a volumetric dataset ready for segmentation.
Step 4 - Segmentation: Threshold-based surface determination identifies the voxel density value that corresponds to the material boundary. This is where a bad decision has outsized downstream consequences. A concrete example of what failure looks like: a polymer housing with a 0.05 mm nominal wall between an internal air channel and the surrounding material has a density gradient across that wall spanning roughly 3-4 voxels at 50 µm resolution. If the threshold is set 5-10% too high - toward the denser side - the algorithm classifies the air-side voxels as solid material, the 0.05 mm wall reads as fully solid, and the channel disappears from the mesh entirely. The CAD model then shows no channel. The part gets machined to a solid geometry and fails on first pressurization. This is caught by running dual-threshold bracketing and reviewing the wall-thickness histogram before locking segmentation - any wall under 0.2 mm triggers a manual review of the local threshold decision.
Step 5 - Mesh extraction: STL or OBJ generated from the isosurface. Typical polygon count 500K-10M triangles depending on part complexity and required surface fidelity.
Step 6 - CAD reconstruction: Mesh imported into Geomagic Design X or SpaceClaim. Feature-based solid modeling begins: extrusions for prismatic geometry, lofts and sweeps for internal channels, revolved cuts for bores and grooves. The deviation color map comparing the finished solid to the mesh confirms model accuracy - the target is ≤0.05 mm RMS fit on critical features.
Step 7 - Deliverable packaging: STEP/IGES native solid, annotated PDF with internal cross-section views, and optional STEP AP242 with PMI GD&T callouts directly on hidden features for import into downstream CMM inspection software.
Parts That Are Perfect Candidates for CT Reverse Engineering
| Part Type | Typical Size | kV Setting | Achievable Voxel Size | Expected CAD Tolerance |
|---|---|---|---|---|
| Hydraulic manifold / valve body | 100-400 mm | 225-320 kV | 100-200 µm | ±0.08-0.15 mm |
| Investment-cast turbine blade | 50-250 mm | 160-225 kV | 50-100 µm | ±0.05-0.1 mm |
| Injection-molded polymer housing | 50-300 mm | 100-160 kV | 50-150 µm | ±0.05-0.1 mm |
| Potted electronic enclosure | 30-150 mm | 100-160 kV | 50-100 µm | ±0.05-0.1 mm |
| Medical device / implant | 5-100 mm | 80-160 kV | 5-50 µm | ±0.01-0.05 mm |
| Legacy cast pump / gearbox housing | 150-600 mm | 320-450 kV | 150-200 µm | ±0.1-0.2 mm |
Hydraulic manifolds and valve bodies are the clearest CT use case. Blind-drilled intersecting galleries, O-ring grooves at the bottom of ports, internal thread forms, and pressure-rated wall sections between adjacent galleries - none of these are reachable by any probe or scanner that requires line of sight. A manifold reverse engineering project that would take 12-20 hours of destructive sectioning and registration takes 6-8 hours of CT processing start to finish, and produces a more accurate result.
Investment-cast turbine components with internal cooling passages present wall thicknesses of 0.5-2 mm between the external airfoil surface and the internal passage. The reverse engineering turbine and compressor blades with internal cooling passages work we do relies heavily on CT for exactly this reason - you cannot characterize a cooling passage from the outside.
Injection-molded polymer housings where you need to re-cut the tool: gate location, sink marks, internal rib geometry, and snap-fit boss dimensions all exist inside the part. CT captures them without destroying the prototype.
Legacy cast parts where only one example survives - pumps, gearbox housings, OEM discontinued components - are the highest-stakes CT candidates. Recovering discontinued and obsolete spare parts through reverse engineering is one of the clearest use cases for CT - it is the only responsible workflow when you cannot afford to damage the only surviving example.
CT Scan vs. Destructive Sectioning: A Direct Cost-Benefit Comparison
| Cost Element | Destructive Sectioning Route | CT Reverse Engineering Route |
|---|---|---|
| Sectioning (wire EDM or saw) | $150-400 per cut × 3-8 cuts = $450-3,200 | None |
| Individual scan sessions | $200-600 per section | $500-2,500 (full scan, all features) |
| Registration and assembly labor | $75-150/hr × 4-8 hrs = $300-1,200 | None (volume is already 3-D) |
| CAD reconstruction labor | $800-2,500 | $800-3,000 |
| Accumulated registration error | ±0.1-0.5 mm | None |
| Risk to part | Part destroyed - irreversible | Part untouched |
| Total project cost | $1,750-7,900 + part replacement risk | $1,300-5,500 |
| Typical turnaround | 5-10 business days | 3-7 business days |
The break-even analysis is straightforward: CT is cost-effective versus destructive sectioning when the part has more than two internal feature zones or when the part has any replacement value above roughly $500. For a hydraulic manifold worth $3,000-15,000 in machining cost, the CT scan fee is a rounding error. For an aerospace casting valued at $50,000-500,000, not using CT is the financially irresponsible choice.
One edge case where destructive sectioning is legitimately appropriate: when you need internal material characterization beyond geometry - confirming grain structure, heat treat response, or inclusion chemistry via metallographic prep. CT’s density-contrast imaging addresses most porosity and void questions (see the next section), but if you need a physical metallurgical section, you are cutting the part regardless, so you might as well scan the sections while they are exposed.
Porosity Detection and Quality Inspection as a Bonus Output
This is the dual-output capability that separates CT from every other reverse engineering workflow. When a legacy casting with no surviving drawings is brought in for reverse engineering, the volume reconstruction process that produces the CAD geometry simultaneously reveals internal casting conditions - at no additional scan cost.
During volume review in VGStudio MAX, the porosity analysis module can flag shrinkage-void clusters near structural features such as bolt bosses - if those regions show multiple voids concentrated near outer walls, that pattern can explain chronic cracking previously attributed to installation error (such as overtorque), pointing instead to a casting fill issue in the original tooling. A redesigned drawing could then relocate such features to avoid the shrinkage zone. That analysis costs nothing extra - it comes from the same dataset already being processed for geometry.
VGStudio MAX produces a formal defect report: pore classification by volume, sphericity, and distance from critical surfaces, exportable as a PDF and as a structured data file. For casting acceptance, this output aligns with ASTM E1814 - Standard Practice for Computed Tomographic (CT) Examination of Castings. If you are using the porosity data for acceptance decisions, the applicable severity framework is ASTM E446, which classifies shrinkage severity on a 1-5 scale: Class 1 (very slight, scattered micro-shrinkage) through Class 5 (severe, large interconnected voids). A quality engineer specifying CT inspection should call out the maximum acceptable E446 class on a per-zone basis - for example, Class 2 maximum in the bolt boss zone, Class 3 acceptable in non-structural wall sections - so the CT lab knows exactly what the pass/fail threshold is before scanning begins. If a legacy part needs a foundry drawing and a quality baseline simultaneously, one CT scan provides both - no second test, no second mobilization.
What to Prepare Before Sending a Part for CT Reverse Engineering
Submitting a part without the right information and packaging guarantees delays, rescan requests, or unusable data. Here is what every submission requires - and the specific reason each item affects the outcome.
Part dimensions and weight - measured, not estimated. Outer envelope (L × W × H in mm) and weight to the nearest 0.5 kg. This determines which scanner envelopes the part (micro-CT for anything under 50 mm max dimension; a Nikon XT H 450 or ZEISS METROTOM 1500 for parts up to ~600 mm; high-energy systems for larger assemblies), sets the shipping class, and tells us whether your part clears the table payload limits. A part that shows up at the lab larger than declared may require rescheduling on a different system.
Packaging protocol. Use foam-over-foam with a minimum 50 mm of foam clearance on all six faces. No loose hardware, fasteners, or secondary parts inside the scan volume - a loose bolt that shifts in transit contacts the part and may cause a surface nick that reads as a real geometric feature on the reconstruction. Double-box any part weighing over 5 kg. If the part has thin walls (under 2 mm), call this out explicitly so the lab can add interior packing that prevents resonance during table rotation.
Material declaration - specific alloy, not just metal type. “Steel” is not sufficient. 4140, 17-4 PH stainless, and H13 tool steel all absorb X-rays differently and require different kV settings. A titanium part submitted as “aluminum” will be under-scanned at 160 kV when it needs 225-320 kV, producing a noisy, artifact-laden volume. A rescan after reshipping adds 5-10 days and the full shipping cost. If you do not know the alloy, say so - density can be estimated from the density histogram after a low-dose scout scan, but that adds half a day to the project.
Feature tolerance requirements - the specific dimensions that drive function. Do not submit a part and say “reverse engineer it.” Tell us: the bore that must achieve a 0.025 mm diameter tolerance for a bearing fit; the gallery wall thickness that must hold 3,000 psi; the channel width that must pass 8 GPM at a given pressure drop. The CAD modeler allocates reconstruction time and deviation check density based on where tolerances are tight. A uniform effort across all features means coarse treatment everywhere - calling out the critical features means tight treatment where it matters and efficient treatment everywhere else.
Existing partial documentation. Even a worn assembly drawing, a supplier spec sheet with one nominal dimension, or a photograph of a disassembled unit helps. It confirms obvious dimensions and catches gross segmentation errors - if the drawing shows a 14 mm bore and the model reads 13.6 mm, we investigate before delivering rather than after.
NDA and IP protocol. Confirm chain-of-custody documentation, confirm the data-destruction policy for raw DICOM files after project completion, and confirm whether export control (ITAR, EAR) applies to the part before shipping. CT labs handle regulated and proprietary parts routinely, but the paperwork must precede the shipment, not follow it.
For a complete pre-submission checklist, see what to include in a reverse engineering quote submission and what to send when requesting a reverse engineering quote. These two references together cover every information element needed to scope and price a CT project accurately.
Deliverables You Should Expect from a CT Reverse Engineering Project
The table below shows the standard package versus optional add-ons. The prose beneath it explains what each deliverable actually contains and why the distinction matters when you are handing files to a machinist, a toolmaker, or an FEA analyst.
| Deliverable | Standard Package | Optional Add-On |
|---|---|---|
| Parametric STEP solid | Included | - |
| Annotated PDF cross-sections | Included | - |
| Deviation color map (PDF) | Included | - |
| Native STL/OBJ mesh | Included | - |
| Porosity/void report (VGStudio MAX) | On request | +$300-600 |
| PMI-annotated STEP AP242 | On request | +$400-800 |
| Native SOLIDWORKS SLDPRT | On request | +$200-400 |
| 2-D DXF drawings (critical features) | On request | +$300-600 per sheet |
Parametric STEP solid (STEP AP214 or AP242): Feature tree intact - not a dumb solid. Internal features modeled as revolved cuts, swept channels, and lofted pockets with construction history visible in SOLIDWORKS, Creo, NX, Fusion 360, or any other MCAD platform that reads STEP. You can modify a bore diameter without rebuilding the part from scratch. This is the deliverable that distinguishes professional CT reverse engineering from mesh-dump CT scanning - the STEP file goes directly into your design environment and behaves like a designed part.
Annotated cross-section drawings (PDF): Orthogonal slices through every critical internal feature with dimensions and tolerances called out. On a hydraulic manifold, that means cross-sections through each gallery intersection, port depth, and wall thickness zone - the equivalent of traditional section view drawings. This is the document your machine shop works from. We generate these in Geomagic Design X or SOLIDWORKS using the CT-derived geometry as the master reference.
Deviation color map: Mesh-to-CAD comparison showing where the solid model tracks the scanned geometry and where it diverges. The target is ≤0.05 mm RMS on critical features; anything exceeding 0.1 mm on a functional feature triggers a modeling review before delivery. The color map is delivered as a PDF with a scale bar and flagged max-deviation locations - it is the quality document that tells you the model was not just estimated.
Native mesh file (STL/OBJ): The CT reconstruction output at full polygon density, useful for FEA mesh generation where internal geometry drives thermal or structural analysis, 3-D printing of verification mockups, or archiving the dimensional state of the as-received part independent of the CAD model.
Rush turnaround note: Standard STEP delivery runs 5-7 business days for a single component of moderate complexity; full PMI package 8-12 business days. Rush 3-day turnarounds are available at most CT labs, but the premium is real: 30-50% added to the scan fee. On a medium aluminum casting where the scan fee is $1,200, that rush upcharge runs $360-600 - budget it explicitly rather than discovering it on the invoice. For multi-component assemblies or parts requiring E446 porosity classification, rush timelines extend to 5 business days minimum because the volume reconstruction and segmentation steps cannot be safely compressed below a threshold without increasing artifact risk.
Also worth reading before you specify deliverables: reverse engineering scan to CAD versus product redesign - the distinction between faithful as-built capture and design intent reconstruction affects what parametric feature structure makes sense.
When CT Scanning Is Not the Right Tool (And What to Use Instead)
Not every part with an internal feature needs CT. The decision tree below is the triage logic for determining the right approach before quoting.
The Three-Question CT Triage
Before committing to CT on any project, ask these three binary questions in order:
Q1: Are all the features you need accessible by line-of-sight or CMM probe?
If yes → structured-light scan + CMM probing is faster and cheaper. CT adds no value.
If no → continue to Q2.Q2: Is the part replaceable, and is CT cost ($1,300-5,500) less than 25% of the part’s replacement value or the cost of a wrong dimension downstream?
If no (part is irreplaceable or CT cost is negligible against risk) → CT is justified regardless of Q3.
If yes (cheap replaceable part) → continue to Q3.Q3: Is the tightest internal tolerance looser than ±0.05 mm?
If yes → CT can meet the spec. Proceed with CT.
If no (tighter than 0.05 mm) → CT captures geometry; plan for CMM validation of the critical features.
Part too large: Most industrial CT systems max out at ~800 mm diameter and ~150 kg. Marine propellers, large turbine casings, and structural weldments over 1 meter are outside this envelope. For those parts, structured-light scanning of all accessible surfaces plus targeted destructive sectioning of representative cross-sections is still the workflow.
All features are line-of-sight: If a bore scope or CMM probe can reach every feature you need, CT adds $500-2,500 in scan cost with no offsetting benefit. Use a handheld scanner or ScanArm with a rotary table.
Budget is the hard constraint: CT scan plus CAD reconstruction at $2,000-5,000 is not justifiable for a $200 plastic bracket with only external geometry. Surface scan at $400-800 is the right tool.
Sub-micron accuracy required: CT cannot match a CMM at ±0.003-0.01 mm. For master reference gauges, tight-tolerance mating features, or calibration artifacts, CT captures the shape but CMM provides the authoritative dimension. Use both: CT for geometry, CMM for critical dimension validation.
Dense multi-material assemblies with similar densities: Two adjacent steel alloys with nearly identical density (e.g., 4140 and 4340) may not segment cleanly - the contrast difference is insufficient to define a clean boundary. For these cases, a test scan to assess segmentation quality before committing full reconstruction budget is the responsible step.
FAQ
Can CT scanning reverse engineer metal parts with thick walls?
Yes, with the right kV selection. Standard industrial CT (225-450 kV, e.g., Nikon XT H 450) handles steel walls up to 50-80 mm. For thicker sections, high-energy systems (1-9 MeV linac CT) extend that range to 300-400 mm of steel equivalent. Accuracy degrades with wall thickness - expect ±0.1-0.3 mm in dense zones versus ±0.05 mm in thinner sections. Always state wall thickness and material when requesting a quote so the lab can confirm capability before you ship.
How long does CT scan reverse engineering take start to finish?
Typical project timeline: 1 day for scan (lab time), 1-2 days for volume reconstruction and segmentation, 2-4 days for feature-based CAD reconstruction, 1 day for QA deviation map and deliverable packaging. Total: 5-8 business days for a single component of moderate complexity. Rush 3-day turnarounds are possible at most labs for a 30-50% scan fee upcharge. Large assemblies or parts requiring PMI annotation add 3-5 days.
What file formats does CT reverse engineering produce?
The primary deliverable is a parametric solid in STEP or IGES - compatible with SOLIDWORKS, Creo, NX, Fusion 360, and any other MCAD platform. You also receive the raw STL/OBJ mesh from CT reconstruction, annotated PDF cross-section drawings, and a deviation color map. Optional add-ons include PMI-annotated STEP AP242, native SOLIDWORKS SLDPRT, and DXF 2-D drawings. Request formats in writing before project kick-off so the modeler sets up the feature tree accordingly.
Is CT scanning reverse engineering the same as CT inspection/metrology?
Same hardware and scan process, different downstream goal. CT metrology (per ISO 10360-11) uses the volumetric dataset to measure features against a known nominal CAD model and produce a GD&T inspection report. CT reverse engineering starts with no nominal model - the goal is to create one from the scan. A single CT scan can serve both purposes: the lab produces the solid model first, then runs a deviation analysis against that model as a self-consistency check, and optionally generates a formal inspection report if traceable metrology documentation is required.
How much does CT scan reverse engineering cost compared to regular 3D scanning?
Surface laser or structured-light scanning of an external part runs $300-800 for a simple component. CT scanning adds $500-2,500 in scan fees on top of that, plus roughly equivalent CAD reconstruction labor. Total CT reverse engineering projects typically land in the $1,300-5,500 range versus $600-2,500 for a surface-only job. The premium is justified when internal features drive function - a hydraulic manifold with undocumented gallery geometry where a wrong dimension means a scrapped machined block worth thousands of dollars.
Can you CT scan an assembled product to reverse engineer internal components without disassembly?
Yes, and multi-material segmentation in VGStudio MAX or Dragonfly isolates individual components by density contrast. The physics that makes this work: density contrast. Steel inserts at ~7.85 g/cm³ are clearly distinct from an aluminum housing at ~2.7 g/cm³, which is clearly distinct from polymer seals at 0.9-1.4 g/cm³. The X-ray attenuation difference between these material classes is large enough that segmentation is straightforward.
The practical limit is contrast-poor boundaries. Epoxy potting compound runs ~1.2 g/cm³, which overlaps significantly with many engineering polymers (ABS at 1.04 g/cm³, nylon at 1.13 g/cm³) and sits close enough to some polymer composites that the threshold line between the potting material and the encapsulated plastic component becomes ambiguous - you are trying to segment two materials with densities within 5-15% of each other rather than the 3× contrast ratio you get with aluminum versus steel. Air at 0.0012 g/cm³ provides strong contrast against almost any solid material, which is why assemblies with air gaps at component interfaces segment cleanly; assemblies where epoxy fully fills those gaps lose that contrast. For any potted assembly, share a BOM with material densities when you submit - it lets segmentation thresholds be tuned before the scan rather than discovering the ambiguity during reconstruction.
Get a Fixed-Price Quote for Your CT Reverse Engineering Project
If you have a part with blind bores, internal galleries, a sealed assembly you cannot disassemble, or a legacy casting with no surviving drawings, CT scanning is almost certainly the right first call. We confirm capability and provide a fixed-price quote within one business day - submit your part dimensions, material, and a description of the internal features you need captured.
Our CT reverse engineering projects start at roughly $1,300 for a small aluminum component and scale from there based on part complexity, required voxel resolution, and deliverable package. Rush 3-day turnarounds are available. ITAR-compliant chain-of-custody is standard.