On 12 November 2026, at MITEX 2026 in the session "Additive technologies for the restoration and modernisation of tooling", we went through how 3D scanning is used in tooling repair. The talk covered wear assessment, preparation for restoration and verification of the result: inside these three stages, four decisions are made on the basis of 3D scanning. Below, the same sequence is set out as an instruction.
Unusual as it may sound, scanning does not speed up the repair itself, but it helps avoid work that would have been done in vain. Some operations it does speed up: capturing the shape of the entire accessible surface with a scanner is faster than obtaining it from point-by-point measurements. Let us go through the four decisions that need geometry, and the limits of the method.
On almost any shop floor the repair chain is built the same way: first the condition of the part, the amount of wear and whether it is worth restoring are established. Then the restoration is prepared, defining the target geometry and the amount of metal to be returned; after manufacturing or restoration, the part is checked for fitness for the next stage.
There are three stages, but four decisions inside them, and each needs its own data: the condition of the part, the target geometry, the allowance and the fitness of the result.
Repair or replace. Geometry provides the input on the extent of wear, the stock available for machining and the body available for weld overlay, but on its own it does not determine the decision. The stock and the body for overlay are calculated against the target geometry and agreed datums; the route also depends on material, defects and technology. Comparison with the cost of replacement remains with the enterprise.
Where to take the target geometry from. This question is often more important than the measurement itself: if the correct geometry is not established, the part can be restored to the wrong shape.
How much metal to add. Weld overlay requires an allowance calculation, otherwise there may not be enough metal, or hours will go into removing the excess.
Is the result fit for use. Inspection after restoration or printing helps determine the next step: accept the part, rework it or make it again.
Two of the four questions concern the geometry of restoration: here scanning gives a distribution of deviations across the surface rather than individual points available from caliper measurements.
Before scanning, agree on what you compare against, what the part is located relative to, and which condition defines acceptance: without that, the deviation map will not answer the question asked.
What we compare against. The actual geometry is compared with the target model or documentation; this reference defines which shape is considered correct.
How we locate the part. The part is located relative to the target model: datums and the coordinate system define which of its surfaces serve as the origin, and the alignment method determines how the part sits on the model. This is what decides which surfaces the deviations are calculated on. If datums are not defined or do not match the part requirements, the deviations are calculated on the wrong surfaces, and the argument stops being about the part and becomes about the method.
Which condition defines acceptance. The acceptance criterion is a number or a condition by which the part is found conforming; the phrase "check the geometry" does not serve as such a criterion.
Geometry is compared not for its own sake but for its role in the way the part works. Working surfaces mate with other parts: they sit on a shaft, guide motion, seal, transmit force. A deviation on them changes the fit - clearance, interference, contact or misalignment - while a surface outside any mating can deviate noticeably and change nothing. That is why requirements name not "the whole part" but the areas whose geometry is responsible for function: on the locating taper of an arbor what matters is how it seats in the spindle and holds the tool, not the shape of a stiffening rib.
The controlled parameters are a list of surfaces and values that the decision depends on; a full scan is useful for further analysis. But the decision is made on the zones named in the requirements: the list is set by the requirements, not by measurement convenience.
If the initial conditions are not agreed, after the colour map is built the question of what to do next remains: the map shows deviation from the model, but not the cause and not conformity. Colour marks the difference from the model, not a diagnosis.
An example from our practice is a drill bit: it is a cutting tool, but the logic of its repair is close to tooling repair, because wear of the working surfaces affects performance. Weld overlay of drill bits is an established practice; service providers report restoring gauge surfaces, the body blades and the areas around the pockets. This concerns the body, not the diamond cutters.
We scanned a bit to assess wear and to see the condition that leads to the bit jamming; jamming stops drilling and, in the best case, leads to losing the tool, and in the worst case, to a change of the drilling route and many hours of downtime. We did not calculate the cost of such an event on a specific well.
Scanning was performed on PRIME: specified accuracy up to 0.014 mm in zone S, 12 MP cameras, the instrument type is approved, reg. No. 92269-24. A datasheet characteristic of the instrument and the result of a specific measurement are different things; what is given here is the datasheet characteristic.
We checked the working profile, the position of the cutters and the symmetry of the blades: these parameters affect how the bit performs in the well, although they do not describe it fully. The workflow included scanning the surface, comparison with the CAD model, building the deviation map and defining the report parameters, where the measured values and the datums are recorded.
The limit of this example: we measured wear but did not perform the weld overlay or the inspection after it; the next stage involves comparing the resulting surface with the target one on agreed datums. Its applicability to a specific part is agreed separately.
Geometry does not answer questions about material, internal defects, the condition of joints, hardness or service life, so other methods are needed to assess them. On accessible surfaces the position of the cutters can be measured, but not their condition or the properties of the material.
The next task is to establish the shape to return to: if a new part is made from a scan of a worn one, the result is a copy of the wear together with the accumulated deviations.
There are four sources of target geometry.
Current documentation or a CAD model. This is the most reliable option if the documentation matches the part; on older equipment the condition has to be checked. Drawings were changed during repairs, and an archived version does not always match the part as manufactured.
A serviceable sample. It works if there is a second part in working order, but it also has to be checked: "serviceable" often means only that the part still works.
Surviving features of the part itself. A worn part may retain areas that never worked: they are used as a source of shape, sometimes as a datum. The suitability of an area for reconstruction and for locating the part is checked separately, because a less worn area does not become a reference automatically.
Engineering reconstruction. If there is neither documentation nor a sample, the shape is restored by calculation and agreed with the enterprise's designer or process engineer; this is a separate job, and its cost is part of the project.
The talk covered two jobs: reconstructing the geometry of an insert from its surviving features, and the forming cavities of a mould from the best preserved surfaces. The scanner and software were ours, the work was done by a partner.
The target model is approved by the enterprise's designer or process engineer; our task is to provide the geometry and show how it differs from the actual shape.
The third task is to determine how much metal has to be returned: weld overlay restoration requires stock for subsequent machining. At the same time, metal above the target surface does not yet mean that the allowance is secured.
At the talk this logic was shown as a training diagram with three states.
A - the allowance is secured, the geometry condition is met; the part goes for machining along the agreed route and subject to the remaining checks.
B - the metal is above the target contour, but the required stock for machining is not secured in every area.
C - the target geometry was not reached, and machining cannot fix it; a different route is needed, for example additional weld overlay; it is determined by the process engineer.
The diagram is a training aid and does not replace a measurement result: in real work the states are determined for specific zones and tolerances. One part can be in different states in different areas.
The allowance is checked against the target geometry and agreed datums, not against the difference between two scans: the difference shows how surfaces changed, but not whether the stock is sufficient. Calculating the volume requires correct alignment, sufficient data and a known initial condition.
That is why the target model, datums, controlled areas and the minimum allowance are agreed before weld overlay; otherwise metal can be added without producing a usable part. The decision to release the part for machining is made by the process engineer, and scanning gives them the distribution of stock over the surface as input data.
The other part of the chain is inspection after additive manufacturing; an example is a milling machine arbor after printing, scanned on PRO II. The specified accuracy of the instrument is up to 0.024 mm in zone S, 6 MP cameras, the instrument type is approved.
The report on this arbor contains two deviation maps, for the small and the large part of it: a map shows deviations, but on its own it is not acceptance. The report provides the basis for a conformity decision; acceptance is performed by the receiving party under its own procedure. For that, the report needs five things:
Repair follows the same approach: the stage changes, the logic stays; the bit and the arbor have their own datums, zones and tolerances. The general order is to agree the requirements, define the location and evaluate the parameters.
The limitation remains as well: the map does not confirm the absence of internal defects and does not determine hardness; other methods are needed for that, and their necessity is best established before acceptance.
Some tasks do not require scanning.
First: if a few simple dimensions have to be checked, conventional measuring tools and a gauge may be simpler and cheaper; the choice depends on access to the surfaces and on the requirements.
Second: conditions may get in the way, because the scanner captures surfaces accessible to the measuring system, but does not see closed zones and channels. Gloss and contamination do not always require matting; preparation, including cleaning and, where necessary, coating, is determined by the part and the requirements. Its time and consumables are part of the estimate.
Third: scanning does not check internal defects, hardness or service life.
Cost is a separate question: the benefit of the method and the decision to buy an instrument are not the same thing. Owning a scanner requires workload, staff, software and process support; a recurring task and an expensive error are a reason to do the calculation, not a ready answer. For a one-off task, a service should be considered.
The economics are calculated over the whole route, not only over scanner time.
Additional costs include surface preparation, scanning, data processing and modelling; at the same time, rework and remanufacture may decrease when real interfaces were not taken into account during preparation. Less excess metal, less weld overlay and less tool wear are also possible, as is the elimination of handing a non-conforming part further down the line.
This is a potential effect, which is confirmed after the fact: by checking which decision actually changed after the measurement.
What has to be compared is the current route and the proposed route with an equally fit result; in-house inspection may already exist at the enterprise, and then it is part of the baseline. If the reduction in operations does not cover the additional costs, scanning is not needed here: that is a normal conclusion, not a failed project. Money, lead times and man-hours are counted separately, without recording the same loss twice.
An example of working conditions, not a promise: in this case we checked positioning without disassembly, the scanner stood next to the tooling, and the line was not stopped. This describes one specific job; it cannot be transferred to all tasks.
If you decide to try 3D scanning for planning or inspecting restoration, six points have to be agreed before the first job:
These points are best collected on one sheet: agreed initial data become the basis for evaluating a pilot, because they show what is being checked and what counts as the result.
Scanning makes sense when geometry affects a decision: a complex surface is hard to verify with individual dimensions, and an error costs more than the measurement. In such conditions it helps reduce rework and prepare restoration more accurately.
If you have a task involving tool inspection, preparation for restoration or verification of additively manufactured parts, tell us what needs to be checked. For a preliminary assessment, a description of the task, a photo, the overall dimensions and the requirements are enough; attach the documentation if you have it. We will say what can be measured and in what form the result can be provided.
