Application of Blue Light Scanner in Turbine Blade Manufacturing
Blades are the parts that decide whether a turbine runs smoothly, and they are hard to measure. This page explains where a blue light scanner fits into blade manufacturing, what it can and cannot hold, and how to read the data it produces. Written for process engineers, quality engineers and buyers who sign off on airfoil geometry.

Why blades are measured with light instead of a touch probe
A blade is a thin, twisted, free-form surface. Measuring it is a different problem from measuring a bracket.
What makes a turbine blade difficult to measure
A blade converts fluid energy into rotation. In an aviation turbine engine, a gas turbine or a steam turbine, the same part does the same job: it sits in a hot, fast stream and turns. What makes it hard to inspect is not the size. It is the shape.
The airfoil is a thin, twisted, free-form surface that changes section from root to tip. Wall thickness on a hollow blade can be 0.8–3 mm. The leading edge radius may be smaller than 0.5 mm. A touch probe with a Ø2 mm stylus cannot reach into the trailing edge without rubbing the surface it is trying to measure.
Contact measurement adds another problem. A CMM produces a few hundred points on a good day. Airfoil form error and chord-wise twist need thousands of points to see the actual surface. That gap is where optical scanning earns its place in blade manufacturing.
How a blue light scanner works and why the wavelength matters
The scanner projects a structured pattern of blue light onto the part and captures it with two or more cameras. Each pixel becomes a 3D point through triangulation. A single frame can produce a million points in under a second.
Blue light sits around 450 nm, which is shorter than white light. On polished metal, that shorter wavelength scatters less and produces a tighter spot, so the point cloud stays cleaner on reflective airfoil surfaces. This is the practical reason blue light scanning is preferred over white light scanning on machined blades.
The equipment still needs help. Bare titanium and polished Inconel reflect too much for any optical scanner. Operators apply a thin removable coating, usually a matte spray or a fine powder, to break up the reflection. The coating is a few micrometres thick and is wiped off afterward. For parts that will later be coated or heat-treated, that step is harmless. For a finished blade with a specific surface finish, it is not.
- 1Structured lightPattern projected and captured, then triangulated into points.
- 2Multiple camerasFill in the shadows that a single camera would leave.
- 3Matte coatingNeeded on polished metal; adds a few micrometres.
Setting up a blade for scanning: alignment is the whole job
A point cloud is useless without a coordinate system. The scanner needs the part in a repeatable position, and the software needs to know where the blade datum is. On a turbine blade, the datum is usually the root platform or the fir-tree root, not the airfoil itself.
The standard approach is to fixture the root in a kinematic mount, scan the whole blade in one or more passes, then align the cloud to the CAD model using the root features. Best-fit alignment alone will hide a twisted airfoil. Constrained alignment on the root, with the airfoil left free, shows the twist. Which one to use depends on what you are trying to prove.
Scan coverage is the next decision. A full scan of a 300 mm blade at high resolution can take 20 to 40 minutes and produce a file in the hundreds of megabytes. If you only need to check one section, scan a band around that section and skip the rest. That is a trade the engineer makes, not the software.
Temperature matters more than people expect. A 300 mm steel blade grows about 0.003 mm per degree Celsius. If the part is scanned at 25 °C and the CAD model is defined at 20 °C, that is a real deviation. For critical airfoils, let the part soak to the reference temperature before scanning.
Matching the measurement method to the blade feature
No single method covers a whole blade. Most production inspection plans use two or three.
| Feature | Best method | Typical resolution | Limitation |
|---|---|---|---|
| Airfoil form and twist | Blue light scanning | 0.02–0.05 mm | Needs matte coating on polished metal |
| Root fir-tree profile | CMM contact | ±0.005 mm | Slow; a few hundred points |
| Leading edge radius | Blue light or optical CMM | 0.01–0.03 mm | Small radius is hard for any stylus |
| Wall thickness (hollow blade) | CT or ultrasonic | 0.05 mm | CT is costly and slow |
| Tip clearance in assembly | Blue light scanning | 0.03–0.08 mm | Needs the blade in its fixture |
| Surface finish Ra | Profilometer | 0.01 μm | Destructive to the position measured |
What the point cloud actually tells you about the process
A color deviation map is the first output, but it is not the answer. The map shows where the part differs from CAD. It does not tell you whether the error came from the casting, the fixture, the tool path or the heat treatment.
Section it. Cut the cloud at five or ten stations along the chord and compare each section to the nominal profile. That is where you see a consistent bias, which usually means a fixture or alignment problem, versus a random scatter, which usually means process capability.
For a machined blade coming off a 5-axis center, the scanner is often used the other way around. We scan the raw forging or casting, align it, and generate the toolpath from the actual stock rather than the nominal model. On a thin airfoil with 1–2 mm of stock variation, that step alone can save a setup. It is the same application of blue light scanner technology that reverse-engineering shops use, just pointed at production.
The scanner is also the fastest way to catch a twisted blade after heat treatment. Distortion from stress relief shows up as a smooth, full-length deviation along the chord. A CMM with 200 points will miss it. A dense cloud will not.
When a blue light scanner is the wrong choice
It is a surface tool. It cannot see inside a hollow blade unless the internal geometry is exposed, and it cannot measure wall thickness through solid metal. For internal cooling channels, CT or ultrasonic is the only option.
Accuracy has a floor. A good blue light scanner holds roughly 0.02 to 0.05 mm on a well-prepared surface, before alignment error. If the print calls for ±0.005 mm on the airfoil, the scanner is not the acceptance tool. It is a screening tool that tells you which parts to send to the CMM.
Reflective, oily or dusty parts need preparation. Every coating step adds handling and a small risk to a finished surface. On a blade that has already been coated, scanning means either accepting the coating thickness as part of the measurement or stripping and recoating it.
Volume matters too. For a 10,000-piece run of small blades, a dedicated fixture and a CMM program will usually beat full-field scanning on cycle time. Scanning pays off on complex geometry, one-off parts and first-article work.
Where scanning sits in our blade and airfoil work
We machine blade fixtures, airfoil gauges, impellers and small turbine components on 5-axis centers to ±0.005 mm and Ra 0.8–1.6 μm. For thin airfoil sections and free-form surfaces, we use scanning to confirm stock condition before cutting and to check form after cutting.
Scanning data feeds back into the CAM setup. If the incoming casting is offset by 0.3 mm, we shift the toolpath instead of cutting into the thin wall. That is the part of the process that prevents scrap, not the report at the end.
Every part ships after 100% inspection. Raw material is checked on receipt, in-process dimensions are monitored during cutting, and final inspection reports are available on request. We work to ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022. Uploads stay confidential and an NDA is available on request.
If you have a blade, vane or airfoil that needs both machining and geometric verification, send the model and the tolerance callouts. We will tell you which features we can hold on the machine and which ones need a scanning or CMM pass.
Common questions
What accuracy can a blue light scanner hold on a turbine blade?
On a prepared surface with a good matte coating, expect 0.02–0.05 mm for form and twist over a 200–300 mm blade.
That figure includes alignment error. If the print tolerance is tighter than 0.02 mm, use the scanner to screen and a CMM or optical CMM to accept.
Does the matte coating damage the blade surface?
No. It is a removable spray or powder a few micrometres thick, wiped off after scanning.
It is not suitable for a blade that already has a final coating or a controlled surface finish, because the handling and cleaning step can affect it.
Can a blue light scanner measure wall thickness on a hollow blade?
No. It only captures external surfaces. Internal cooling channels and wall thickness need CT or ultrasonic testing.
Many shops combine both: scan the outside for form, CT a sample for wall thickness.
How long does scanning one blade take?
A full high-resolution scan of a 300 mm blade runs 20–40 minutes, plus setup and alignment.
If only one or two sections are needed, a band scan cuts that to a few minutes per part. The right choice depends on what you are checking.
Can scanning replace a CMM for first-article inspection?
Not on its own. Scanning gives dense coverage of form and twist, which a CMM cannot match on a free-form surface.
For tight tolerances and datum-controlled dimensions, the CMM is still the acceptance tool. Most first-article plans use both.
Can you machine a blade from a scanned point cloud?
Yes. When no CAD model exists, we scan the part, align and clean the mesh, and build a machining model from it.
Where a nominal model does exist, we scan the incoming stock, align it, and adjust the toolpath to the actual material condition.
Send us a blade or airfoil model
Tell us the tolerance callouts and the feature list. We will come back with a quote, a DFM note on which features need scanning or CMM, and a machining plan.
Quote within 12 hours±0.005 mm machining100% inspectionNDA on request