Industrial CT Buying Guide: Prove the Required Part Decision
An industrial CT purchase should prove that the proposed system can support your required decision on your parts. A small voxel specification, a polished volume rendering and a dimensional result each answer different questions.
For: U.S. NDT equipment coordinators, quality staff and inspection business owners
State the decision the CT result must support
Start with the question the organization needs to answer. Locating an internal feature, evaluating a discontinuity, comparing a surface with a drawing and measuring an inaccessible dimension are different objectives. Ask the technical owner to write the required output and the evidence needed to accept it. Purchasing can then compare systems against that defined task instead of treating every three-dimensional image as equivalent capability.
Create a part-family brief with material, overall geometry, region of interest, required output and representative variation. Include what the organization already knows from other methods and what remains uncertain. This guide concerns acquisition decisions and evidence management. It does not prescribe radiation settings, reconstruction parameters, safety arrangements or inspection acceptance limits. Those belong to the responsible technical and site processes. A supplier demonstration should support those processes with evidence for a specific configuration and task.
Separate three meanings of resolution
A voxel describes a volume element in the reconstructed data grid. Its dimensions help describe the dataset, but a small grid spacing alone does not prove that a particular feature is reliably distinguishable. The University of Texas CT laboratory explains that image detail depends on several physical and geometric factors. Its resolution discussion also distinguishes spatial detail from the ability to distinguish differences in attenuation. Treat the page as an explanation of principles, not a current equipment specification.
Dimensional measurement introduces another question: how well does a reported dimension represent the intended measurand? NIST's CT traceability research discusses measurement uncertainty and scale calibration. Ask the supplier to identify whether a claim concerns grid spacing, demonstrated image detail or dimensional measurement performance. Do not place those claims in one comparison column called accuracy. The distinction prevents a visually impressive demonstration from being mistaken for evidence supporting a tight dimensional decision.
Choose trial parts that represent the difficult work
Select representative parts through the technical owner, including the material and geometry combinations that drive the purchase. A small, simple demonstration object may be useful for introduction but cannot stand in for a substantially different production part. Record why each trial part was chosen. Include relevant variation rather than testing only the easiest specimen and assuming the rest of the family will behave identically.
Give every trial item a unique identifier and retain its condition and provenance. If a part contains known features, distinguish independently established information from assumptions supplied with it. Define which information the supplier may use during the trial and which results will be compared afterward. This helps the organization evaluate the evidence fairly. It also prevents a later reviewer from confusing a result generated with extensive prior knowledge with a demonstration of an unknown-part examination workflow.
Evaluate usable part capacity rather than cabinet size
Ask the supplier to distinguish the physical space available inside the system from the configurations that support the required image or dimensional task. The part, fixture and required movement all matter. A component fitting through the door does not establish that its region of interest can be examined as intended. The University of Texas discussion also identifies material attenuation as a constraint separate from physical accommodation; avoid treating the largest listed object size as a universal material capability.
Request a documented configuration for each representative part family. Record any restrictions involving fixtures, orientation, access to a region or optional system components. The technical owner determines whether the proposed arrangement is suitable. Purchasing should retain the configuration and make sure its necessary components are included. If different tasks require materially different arrangements, compare those as separate supported cases instead of merging their best specifications into a fictional single configuration.
Distinguish an attractive rendering from the retained evidence
A rendered surface can communicate a result clearly while omitting information available in the reconstructed volume or acquisition data. Ask which datasets the supplier retains, which are required for the intended review and which are created only for presentation. Do not assume that receiving a rotating three-dimensional video gives the organization the ability to revisit the underlying analysis. Define the required deliverables before the demonstration begins.
Have a reviewer identify the relationship between projections, reconstructed volume, derived surface, measurements and final report where those artifacts are part of the agreed workflow. Record the software and versions used to create each. The organization need not retain every intermediate file by default, but its retention choice should be deliberate and consistent with the technical and customer requirements. Ask what future questions cannot be answered from the proposed retained package so the storage decision has a clear consequence.
Require task-specific dimensional evidence
For a dimensional application, ask what supports the reported result for the intended material, geometry and feature. A general instrument specification may not describe every measurement task. NIST's research paper addresses the relationship between instrument calibration, uncertainty and traceability. Use that distinction to request an explanation of the evidence for the proposed measurement, rather than accepting a blanket claim that every CT dimension is traceable because the equipment has a certificate.
The metrology owner should determine the relevant measurand, comparison artifacts, uncertainty requirements and acceptance approach. Purchasing records that decision and obtains the agreed evidence. Do not invent a universal ratio between voxel size and acceptable dimensional tolerance. When the supplier proposes a demonstration against another measurement method, document the reference information and its limitations as well. A numerical difference is difficult to interpret if nobody knows how the comparison value was established or what uncertainty accompanies it.
Preserve segmentation and analysis choices
Ask how the software documents the steps that turn reconstructed data into a surface, a feature assessment or a dimension. The buyer does not need to select analysis settings in the request for quotation. The technical reviewer does need enough information to understand what produced the reported result. Identify which choices are saved with the project and which exist only in an operator's session or notes.
Include a repeat-review exercise in the trial. Another authorized reviewer should open the retained project and locate the relevant result using the supplied documentation. Record whether the software preserves the required context and whether additional manual explanation is necessary. This is not a claim that different analysts must always produce identical results. It is a check that the proposed record allows the organization to review the analysis, understand relevant choices and manage authorized changes without reconstructing the entire history from memory.
Measure complete turnaround on the representative part
Define the start and finish of the trial's time measurement. Loading, identification, acquisition, reconstruction, analysis, review and customer delivery can each affect turnaround. Record the stages relevant to the intended business process. A supplier's scan duration may be accurate for its stated conditions while representing only part of the time needed to deliver an accepted result. Ask for the complete workflow to be demonstrated on a representative part.
Note whether acquisition and analysis can occur concurrently on the quoted hardware and licenses. If a demonstration uses an additional workstation or specialist service, include that dependency in the comparison. Identify any manual preparation of reports or exports. Use measured trial observations and explicit workload assumptions in the business case. Do not translate the shortest demonstration time into an unconditional annual capacity promise, especially when the proposed part mix includes materially different sizes, materials or analysis requirements.
Test the actual analysis workstation and reader access
CT data can place substantial demands on the review workflow, but a generic computer specification does not establish a usable experience for a particular dataset. Ask the supplier to demonstrate the intended project on the proposed workstation or a documented equivalent. Include opening, navigation, analysis and export tasks that the reviewer actually performs. Record dataset size and software version so the result remains meaningful when comparing offers.
Separate acquisition, reconstruction, analysis and read-only access rights. Determine whether the organization can review retained projects when the scanner is occupied, serviced or disconnected from a particular account. Ask what happens when a license expires and what supported reader remains available. These are commercial and records questions as well as usability questions. The final package should provide the required access for the retention period established by the organization, without depending on an undocumented temporary demonstration license.
Build a sample evidence package before signing
Request a complete package for one fictional job containing the agreed datasets, report, identity manifest and relevant configuration information. Give it to the intended receiving reviewer. Ask the reviewer to identify the part, locate the required result and explain the limits of the evidence. If the recipient needs proprietary software, confirm the license and installation arrangements rather than assuming an exported file is self-contained.
Include a controlled recovery exercise for retained test data. Restore the package through the organization's intended storage process and repeat the review. Note any missing linked files, external references or database dependencies. A project that opens correctly on the supplier's demonstration computer may rely on resources not included in the delivered folder. The acceptance record should show that the organization can recover and interpret the agreed package in its own supported environment.
Work through a fictional voxel-size mismatch
A hypothetical manufacturer requests a CT system to measure an internal feature in a dense assembly. One quotation emphasizes a very small minimum voxel size demonstrated on a much smaller object. The buyer asks whether that specification applies to the actual assembly and the required measurement task. The supplier explains the configuration difference and proposes a representative trial. The buyer does not convert the brochure number directly into an acceptance tolerance.
During the trial, the technical team evaluates the intended result and its supporting evidence, including the limitations of the comparison reference. It may accept the proposed system for feature location while requiring different evidence for dimensional acceptance. The purchase scope is adjusted accordingly. This fictional example illustrates why imaging capability and dimensional authority should be stated separately. A useful system can support one defined task without justifying every measurement claim that might be made from its images.
Compare buying equipment with purchasing a defined service
For an intermittent or still-developing application, an external CT service may help establish the required output before the organization selects equipment. Ask for the same clarity about part scope, retained data, analysis access and result limitations. A service trial should produce evidence the organization can use to define its requirements, not only a promotional image. Preserve the configuration and assumptions so the result can inform a later purchase comparison.
When comparing service and ownership, include the activities that remain with the organization in both cases: defining the task, controlling part identity, reviewing the result and retaining the accepted record. For ownership, add the actual facility, software, staffing and support dependencies identified by the project team. Avoid presenting a universal break-even volume. The meaningful decision depends on the organization's workload, turnaround needs and supported application scope, not merely the price of a scan divided into the scanner price.
Define factory and site acceptance responsibilities
Agree which evidence will be generated before delivery and which must be established after installation. A factory demonstration can show a configured capability, while site acceptance addresses the delivered system in its intended environment and workflow. The technical, facility and safety owners determine the necessary checks. Purchasing should identify the evidence, responsibility and disposition process for each item without attempting to replace those specialist decisions with a generic checklist.
Keep a list of departures from the demonstrated configuration. A different workstation, fixture, software revision or analysis package may require review before the prior evidence is treated as applicable. Record who decides whether a change is material. Make unresolved items visible in the acceptance decision and payment milestones as appropriate to the agreed contract. This gives both parties a clear route to close the purchase without implying that shipment alone establishes the complete intended capability.
Maintain the asset record alongside the technical evidence
Retain the final part-family scope, trial results, accepted configuration, software entitlements and record-delivery instructions in the controlled purchasing file. Keep the boundaries of the accepted applications visible for future users. A later expansion to a different material or measurement question should be reviewed through the appropriate technical process rather than inferred from the scanner's availability.
NDT Connect can record the scanner and associated equipment by identity, due date, status and notes. Its equipment tracker does not reconstruct CT data, calculate uncertainty, store volume datasets or approve dimensional results. Use it for coordination and references to the controlled evidence where appropriate. The completed purchase should leave the organization able to explain what decision the system supports, which evidence establishes that scope and how the resulting record will remain reviewable.
Record what the trial did not establish
Add an explicit exclusions field to the trial report. It might state that a particular part family, region of interest or dimensional characteristic was not evaluated. This is more informative than a broad successful demonstration label. Have the technical owner review those exclusions before the commercial decision, because an untested requirement may be central to the original business case.
Preserve unsuccessful and inconclusive trial observations with their disposition. They can explain why the final configuration differs from the initial offer and prevent a future buyer from repeating an assumption already investigated. The record should distinguish a demonstrated limitation, an incomplete trial and a capability that was simply outside the agreed test scope.
Frequently Asked Questions
Does minimum voxel size establish dimensional accuracy?
No. Voxel dimensions, demonstrated image detail and uncertainty of a particular dimensional result are different properties.
Is fitting a part inside the cabinet enough?
No. Evaluate the supported imaging configuration, material and required region of interest for the actual task.
What should a CT purchase trial deliver?
A representative result plus the agreed underlying records, configuration, analysis context and evidence needed to assess the intended application.
References & Standards Cited
- NIST: dimensional measurement traceability by X-ray computed tomography View source ↗
- University of Texas CT Laboratory: resolution and size limitations View source ↗
- NIST-hosted paper: CT dimensional measurement traceability View source ↗
Related on NDT Connect
Join NDT Connect — free
The free marketplace for the NDT industry — connect inspectors and the companies that need them.
Free to join · Compare provider qualifications and confirm availability directly · Browse providers
By NDT Connect Editorial · Updated
Source-based editorial guidance. Read our editorial policy and disclosures.