Asset Integrity Digital Twin — Your Inspection Data and CMLs on One 3D Model

Asset integrity digital twin for inspection: the direct answer

An asset integrity digital twin for inspection is a 3D model of your fixed equipment where every inspection result sits at the spot it describes. A thickness reading belongs to a CML, an MFL indication to a floor plate, a weld result to a weld number. The signed report stays attached. Inspectors and turnaround planners read condition by location and by campaign instead of opening folders of PDFs. The owner's inspector still makes every code decision.

What an inspection-data twin is, and what it is not

"Digital twin" covers very different products. A process twin simulates how a unit runs. A reality or BIM twin stores geometry and a point cloud. An inspection-data twin stores condition evidence: what was measured, where, when, by whom, with which instrument, and what the reviewer concluded. For tanks, vessels and piping, that evidence is what tells you how the metal is degrading.

Three quick tests tell you which kind of twin you are looking at:

• The location test. Click a point on the model. You should get a reading tied to a CML, TML or weld ID. If you get a list of files, you have a viewer with attachments.
• The history test. Pick one CML. You should see its readings from every campaign side by side, without opening each report.
• The record test. From any reading, you should reach the approved report, with the technician, reviewer and approver named.

A viewer with attachments is still useful for orientation. It will not help an inspector review corrosion trends or help a planner build turnaround scope, because the data cannot be compared by location.

Be clear about the limits too. The twin is not the inspector of record. It does not set API 510, 570 or 653 intervals, and it does not run risk-based inspection or fitness-for-service assessments. Atlantis does not offer RBI or FFS. The twin holds the measured, traceable data that the owner's inspector and engineers use for those decisions.

Which data feeds an inspection twin

A twin is only as good as the records loaded into it. Most of the data already exists at your site. It is spread across the EAM, inspection files, contractor reports and drawings. The table shows the usual sources and who should own each one.

Two rules keep the data layer trustworthy. First, every reading carries its provenance: method, date, technician and certification level, instrument serial number, probe, calibration block and procedure revision. Second, the twin shows only approved results. Drafts and unreviewed readings stay out until the reviewer signs them off.

How a twin is set up, step by step

The setup work is mostly data work. The 3D part is the easy bit. A typical sequence looks like this:

1. Choose a bounded first scope. One tank farm, one process unit or one turnaround package. A plant-wide first load usually stalls on data cleanup.
2. Use the EAM's equipment numbers unchanged. Twin objects should carry the same tags and functional locations as your work orders. Never invent a second numbering scheme.
3. Reconcile the CML register. Map historical readings to stable CML IDs. Flag points that were renumbered, abandoned or duplicated between campaigns. This step sets the schedule, not the model.
4. Build the geometry. Use entered dimensions for standard shapes such as tanks, shells, heads and pipe runs. Use existing drawings or CAD where they match the field. Commission a laser scan where neither exists, for example through tank farm 3D scanning.
5. Load history with its original dates. Each historical reading keeps its inspection date and source report. Readings that cannot be traced are marked unverified, not loaded silently.
6. Set display rules with the owner's inspector. Color bands, alert levels and minimum thickness values come from the owner's inspector or engineer. They should never be defaults chosen by a software vendor.
7. Move field capture to location-keyed reports. From now on, technicians report against the same CML and weld IDs that the model uses. Offline capture matters at tank farms and in congested units.
8. Agree access and integration. Decide who sees which assets. Decide which records flow to the EAM through the API. Integrations are scoped per project.

The order matters. Teams that start at step 4 get an impressive model with an empty or untrustworthy data layer. Teams that start at step 3 get a plain model that answers real questions on day one.

If you want to see this working on one of your own assets, see Digital Twin reporting with a sample of your data.

How the twin sits next to APM, EAM and the historian

Platform comparisons usually list what each system stores. The practical question is how data moves between them. The rule that prevents most problems: each record type has exactly one system of record, and the others read from it.

• EAM or CMMS to the twin. Equipment numbers, functional locations and work order numbers. The twin uses them. It does not create its own.
• Twin to the EAM. Findings and inspection recommendations that need maintenance work. The EAM raises, schedules and closes the work order. The twin shows the finding and links to the order.
• Historian to the twin. Operating context, such as a temperature excursion or a change of feed, as background when a corrosion trend changes. The historian stays the system of record for process data.
• APM and the twin. Rotating-equipment health, such as vibration on a pump, normally stays in APM. Fixed-equipment degradation is measured by NDE, so it belongs on the inspection twin.
• An existing inspection database. If you already run one, the twin can be the visual layer over it rather than a second database holding the same readings.

Duplicated records are the most common way twin projects lose trust. If a thickness reading lives in two systems, sooner or later they disagree. Nobody then knows which one the inspector used. For plain definitions of each system class, read digital twin vs APM vs EAM vs historian.

Digital twins for turnarounds

Turnarounds put the most pressure on inspection data. The same few people need the same answers, quickly, while the unit is down. An inspection twin helps at three stages.

Before the turnaround

Scope review happens on the model. Planners and inspectors can see which CMLs are due, which recommendations are still open, which welds were rejected last time and which areas were flagged for corrosion under insulation. Method allocation follows from that. MFL goes on tank floors coming out of service. Corrosion mapping goes on the areas where spot readings scatter. Weld NDE goes where repairs are planned. Access is planned on the same geometry. See how scaffold and access planning works in turnaround access planned in the model.

During the turnaround

The model becomes a status board. Each item shows one of a few states: not started, inspected and awaiting review, accepted, rejected, repaired, re-examined. Engineers who are not on site can review approved results the same day. The question "has anyone looked at this nozzle yet?" gets answered by looking, not by phone calls.

After the turnaround

The new campaign becomes the baseline for the next one. Open findings go to the EAM as work. The handover package is the set of approved reports, already placed on the asset. For the whole sequence, before, during and after the shutdown, read digital twins for turnarounds.

We do not publish savings percentages for turnarounds. The benefit depends on how clean your CML register is and how much of your scope is inspection-driven. A pilot on one unit is the honest way to measure it.

Tank inspection digital twin: what goes on the model

Storage tanks suit the twin approach well. Tank geometry is regular, and API 653 inspections produce large, location-specific datasets that are hard to read in tables. A tank twin typically carries:

• Shell courses with UT readings at fixed CMLs, plus corrosion-mapping grids on the lower course and around nozzles.
• The floor plate layout with MFL indications and the UT prove-up depth at each flagged location.
• The annular ring and critical zone near the shell, shown separately because it is evaluated separately.
• Settlement survey points around the shell and across the bottom, by survey date.
• Nozzles, roof and appurtenances, with VT findings and photos.
• Repairs, such as replaced plates, patch plates and re-welded seams, with their examination records.

Each internal inspection adds a new layer on the same geometry. The owner's API 653 inspector reviews the data and makes the interval and repair decisions. The twin makes that review faster to navigate. It does not make the decisions. For tank-specific detail, see the storage tank digital twin page. For the NDE scope that produces the data, see aboveground storage tank inspection.

Who it is for, and where Atlantis fits

Who an inspection twin is for

• Refinery, terminal and chemical plant integrity teams who already hold years of thickness data and cannot trend it easily.
• Mid-size operators with tank farms who need a clear condition picture without an enterprise APM program.
• Multi-site owners who want one CML convention and one view across plants and contractors.
• NDT service providers who want to hand over data that drops straight onto the client's model.

It is not a good fit yet if nobody at the site owns the CML register, or if the real need is process simulation. Fix the register first, or buy a process twin.

Where Atlantis fits

Atlantis works on both sides of the twin. On the data side, our ASNT-certified technicians perform NDE under ASNT Level III oversight: UT thickness at CMLs, corrosion mapping, PAUT, TOFD, tank floor MFL, guided-wave screening, and MT, PT and VT. Radiography runs through licensed radiography crews. On API 510, 570 and 653 work, we deliver the NDE to your Authorized Inspector, who stays inspector of record.

On the software side, Digital Twin reporting places inspection reports and CML data on the 3D asset. The NDT Reports app in Atlantis ERP is where field reports are captured offline, reviewed and approved before they reach the model. Atlantis has delivered 1,500+ inspection activities. We work from Houston and Hyderabad, and US projects are delivered onsite or remotely.

For regional context, see the digital twin pages for Texas and the Gulf Coast, Louisiana, California and the United States. When you are ready, see Digital Twin reporting on your own asset. We reply within one business day.

Choosing a vendor and avoiding common mistakes

How to compare digital twin vendors for inspection

Most vendor demos show the 3D view. Ask about the data layer instead. These questions separate inspection twins from viewers.

Common mistakes on inspection twin projects

• Scanning the whole plant first. Geometry is the cheapest data to replace and the slowest to use. Start where inspection data is richest.
• Renumbering CMLs during migration. New IDs break trending against every earlier campaign.
• Loading scanned PDFs as data. A number in an image cannot be trended. Transcribe and verify, or leave it as an attachment.
• Mixing nominal and measured thickness. Mark which is which, or colors will mislead.
• Treating colors as acceptance. A color is a display rule. Acceptance belongs to the inspector of record.
• No plan for the next campaign. If field reports keep arriving as free-text PDFs, the twin goes stale after the first load.

Frequently asked questions

What is digital twin inspection software?

Digital twin inspection software places inspection results on a 3D model of the asset, keyed to CMLs, welds and floor plates. Each result links to its approved report. Its job is to make inspection history easy to find, compare and share by location. Code decisions stay with the inspector.

Can a tank inspection digital twin hold MFL floor scans and shell UT?

Yes. The floor plate layout carries MFL indications and UT prove-up depths. The shell carries CML readings and corrosion-mapping grids by course. Settlement points and repairs sit on the same model, campaign by campaign.

How are digital twins used for turnarounds?

Before the shutdown, they support scope review and access planning. During it, they act as a live status board for inspected, rejected and repaired items. Afterwards, the new results become the baseline for the next campaign and open findings go to the EAM.

Who decides the color thresholds on the model?

The owner's inspector or integrity engineer, asset by asset. Colors are display rules, not acceptance decisions. A good platform records who set each rule and when.

Does Atlantis act as our API inspector or run RBI in the twin?

No. Atlantis performs NDE and delivers it to your API Authorized Inspector, who stays inspector of record. Atlantis does not offer RBI or fitness-for-service assessments. The twin holds the measured data that your engineers or chosen specialists use for them.

Can our NDT contractor deliver data straight into the twin?

Yes, if the contractor reports against your CML and weld IDs in a structured format. Give contractors the ID list and datum conventions before mobilization, and accept only location-keyed reports.

Is there a digital twin demo we can try?

Yes. The public Digital Twin Reporter at dt.atlantisndt.com has a demo mode with sample data. Demo values are illustrative and are not engineering acceptance criteria. For a walkthrough with your own data, see Digital Twin reporting in a live session.

How should we start?

Pick one tank farm or one unit with good inspection history. Reconcile its CML register, load it and test it against real questions from your inspectors. Expand only once the first scope is trusted.