Thursday, April 21, 2016

CMM inspection of Turbine blade profiles – Past, Present and Future

Most turbine blades are set up and dimensioned in a very traditional way suited to when they were inspected with guillotine style gages. Iterative alignments are complicated to say the least, not just best fitting 6 points in space but sometimes using mid-points or high point local scans to generate one of the best fitting points.

CMM’s have been used to do this inspection instead of dedicated gages in the last decades for a number of reasons;
  1. CMMs are flexible and, once programmed, can measure many different parts
  2. Maintenance, storage and calibration of one dedicated gage per part could be laborious and expensive
  3. Most importantly the industry demanded digital data for every measurement. GO/NOGO or subjective judgements in inspection have become unacceptable. Digital results can also be archived away easily for future reference if needed.
Those of us that have been in the business (measuring blades) for a while are aware of the problems of probing points on profiles such as blades and are very aware of phrases like “tip radius error” and “out of plane error”.  CMM software vendors and CMM programmers have found elaborate work-arounds to compensate these know error conditions over the years, but as 3-axis or even 5-axis continuous contact scanning has become more the norm, then the work-arounds needed to become even more complex to cope with the amount of data produced and the fact that scanning probes have a habit of wandering “out of plane” when the actual surface is away from nominal.

Please forgive my drawing,
It is many years since my drafting days were over! However, it will illustrate the issues.

Remember the profile is to be measured in section planes cut through the airfoil at various heights. The amount of twist and bend in blades produced today also compounds these effects.

The circle represent the CMM stylus ball and the point we really want to measure is Point C and then compare it to the nominal Point D.

Let’s keep it simple and imagine measuring this with a touch-trigger probe. We have 2 possible approaches;

  1. We can move in 2D and approach point D (from left to right), but instead of hitting point D we strike the surface earlier and actually record Point E, giving tip radius error of “X”. This effect can be lessened by using the smallest ball possible, but nevertheless it will still be there
  2. We approach the point along Vector F, but if the actual surface is away from nominal (as in this case), we overshoot and record point B, giving a tip radius error of “X1”. Actually we record Point A and apply the tip compensation in the direction of Vector F and record point B.

If you are scanning the section the similar effects occur. If the surface ‘wanders’ away from nominal, the scanning probe senses a drop in stylus force and vectors towards the surface to get back to its nominal stylus force. Without any further calculations or corrections the Z-axis value will drift, particularly with twisted and bent blade shapes.

As I mentioned previously, it is not beyond a skilled CMM programmer to apply some trigonometry calculations to his program to “approximate” Point C from the measured Point B.

With the Renishaw REVO 5-axis scanning solutions these problems and approximated solutions still occur, but other approaches become possible like sweep scanning.

Software programs developed for this approach program the collection of the sweep scan data (recording stylus ball center data to remove the tip radius error problem) and then intersect the point clouds collected with the section planes. The number of point ‘landing’ on the section plane exactly can be low and/or inconsistent. Software routines then process the point cloud further to push points close to the section plane onto the plane itself. This approach speeds up the process a lot and solves the problems associated with tip radius error and out of plane measurements, albeit with a fair amount of calculation and approximation.

This approach can also be adopted with laser scanners, but the surface reflectivity of the blade can determine the success or failure of the method. Even moderately shiny surfaces can cause issues (and measuring uncertainty) and spraying such parts with powder to fix the reflectivity problem, introduces another problem of powder thickness and consistency of powder thickness, again introducing an unpredictable uncertainty of measurement.

Another area of concern with contact CMM data is the errors generated when measuring the trailing
edge (and for thin blades also the leading edge) of the blade. In this area the change in direction of the surface is rapid and particularly when the profile is a long way from nominal shape, the tip radius correction becomes erratic.

 The error plot sketched on the right is very typical and most people associated with CMM report from turbines blades will be familiar with it.

This area of the blade is perhaps the hardest to manufacture to nominal, so it can be confusing with such reports to know whether this is a manufacturing error or a tip compensation error on the CMM.

The methodology applied with 5-axis scanning on REVO described above does address this issue as well, within the constraints also described.

The sweep scanning methods also produces a massive amount of data describing the complete shape of the blade which can be incredibly useful to engineering, even if it is not giving the inspection group what they need without going through the sectioning process.

Here at Wenzel we have the same issues and work-arounds with our CMM solutions and we have to say that for many parts the CMM is the best solution, particularly when fitted with the REVO 5-axis scanning head. This is absolutely the best solution for blisks, multi-vane segment and large, heavy vanes.

However, we believe we have a better solution for single blades – any blade from 1” tall to 80” tall fan blades. This solution is
a high speed optical scanning machine called CORE. The CORE machine is a 5 or 6 axis optical machine. The sensor, developed by Wenzel, projects a high intensity white spot onto the part. Then one or both cameras in the sensor detect the point of highest light intensity in the middle of the spot. The software then calculates the point position as with any other CMM. The key differences with CORE when compared with other optical systems is that it can operate on even highly polished surfaces and can cope with the finish of any turbine blade without spraying. 

The CORE can also measure highly polished artefacts (like spheres and gage blocks) and so can run standard ISO 10360 tests for CMM verification. 

So whilst the CORE operates like a CMM in picking up the iterative alignment and scanning the profile sections, the key difference for CORE is that the sensor has effectively a zero tip radius and therefore none of the problems documented above occur.

CORE has the added benefits of being shop hardened, occupies a small footprint compared to traditional CMMS and measures blades in at least half the time of any other CMM.

So in conclusion, turbine blades profiles have been measured with CMMs for many years and the inherent problems of “tip radius compensation” and “out of plane measurement” have been worked around. Sweep scanning systems go a long way to correcting these problems, but not without a fair amount of point manipulation and correction to report section scans with enough data.

If you need a machine that accurately measures and reports actual profiles without tip compensation errors and subsequent corrections, that can be certified to ISO standards like a traditional CMM and can measure today’s complex blade shapes way faster than any other measuring machine right out on the shop floor, then perhaps you should take a look at the CORE range from Wenzel.

For a more detailed comparison between bridge CMM and CORE please click here to read a blog from last year.

Andy Woodward

Wenzel America

Big Data – Driving Change in Manufacturing

If you spend any time at all online reading business articles, you’ve likely noticed that we are in the midst of a technology shift which is quickly and fundamentally changing our manufacturing world. 

We are entering an era of massive amounts of data, being brought together by the growing connectivity of our infrastructure, often referred to as the Internet of Things. 

This transition, coupled with increasing automation in the workplace, creates mountains of data which should give us the ability to see into our processes, products, and our companies as whole, transforming how we do business. 

But will it really? Will all this data make our collective lightbulbs go off, helping us to manage and improve our processes in the most efficient way possible? That should be the goal of course, but as with any new paradigm, it can be easy to miss the mark. 

Take CMM data for example. For over 50 years, there have been CMMs available in Quality labs all over the world, bought with the promise of improving the parts being made. But simply inspecting parts, collecting piles of data, and labeling items good or bad, doesn’t really make the most of the investment that has been made in the tool. 

A CMM is most valuable when the data you collect, can drive improvements in your process. The results from inspection should feed decisions in manufacturing to reduce downtime, improve throughput, and efficiency. These things will likely lead to better profits, and customer satisfaction. How do you get there?

With Quality Inspection data the answers are all around us. In the automotive industry, AIAG provides resources for building and maintaining a Quality program that uses statistical controls to help you make decisions; in other words, driving your process with your data.

If you make Medical devices, the FDA sets out numerous guidelines not only for Quality, but also relating good manufacturing processes. 

Not to be left out, the Aerospace Industry is guided by SAE International and their AS9100 guidelines, providing all manner of guidance from CAD data integrity, to standards for reporting results.

The IoT data available to us now, is a bit different though from the topics above. There isn’t necessarily a white paper or industry standard on how to manage this information and make decisions from it, but that doesn’t mean the old ways don’t work. If we step back and look at it from a broader perspective, data is data. The basic tenets of the guidelines from each organization above, have some very similar structure, regardless of the product or process. The main principles are:

  1. Understand the key characteristics of your product or process and plan accordingly
  2. Use an effective tool to monitor these items
  3. Ensure you monitor with enough frequency that you can trust your data
  4. Make decisions with the data
  5. Monitor the outcome
  6. Adjust as needed

If we take this broad view to the topic, we can then apply this philosophy to any data we now have available throughout our organization. Simple right? The problem is, in this modern era, the data is coming faster than ever before, and is so readily available that we often skip steps in the planning process, assuming that the speed and quantity of data will make careful planning a thing of the past. This strategy however, typically leads to less answers than one started with, and the typical reaction is then to scrap it all because the outcome isn’t what was expected. This is often when you hear things like, “doesn’t work in OUR plant…”.

To avoid this outcome and position your company for success in this new data minefield, the planning portion of the process I outlined above becomes even more critical. Understanding your process; what drives it, the key inputs and variables, and what you want to gain from monitoring, are all of paramount importance. 

For example, focus your energy on those items that affect the end product; fit, form, or function. Use accounting or finance data to look for areas of high cost in your process where more monitoring might help you eliminate waste. How does the temperature and humidity in the plant affect your quality? Or the staff?  

It’s really no different than checking parts on your CMM. You wouldn’t waste time inspecting features on the part that have no bearing on the performance, or the customer’s quality. In fact you probably take the data you do gather, and look for ways to reduce cost in manufacturing, or reduce inspections on items that aren’t critical; all to save time and money. 

When we turn that spotlight on our broader processes, the intent should the same. But the questions we can ask in this connected age are limitless and managing metrics can be a full time job. Driving the charts in an upward trend can easily become an obsession that feels like an end in itself. But the data is only valuable if it drives improvement in your business; making your products and processes, better, faster, and more efficient. 

But let’s not forget the skills and techniques that got you to where you are. The basic problem solving principles that you use daily to ensure product compliance, can easily translate in this new paradigm, and can help you drive improvements throughout your organizations. 



RAPID is the Authority on 3D: Printing, Scanning, and Additive Manufacturing. If you are interested in 3D Printing, then it is the one event you need to attend. Why is 3D Printing important?

According to Wohlers Report 2016, the additive manufacturing (AM) industry, consisting of all AM products and services worldwide, grew 25.9% (CAGR) to $5.165 billion in 2015. 

The CAGR for the previous three years was 33.8%. Over the past 27 years, the CAGR for the industry is an impressive 26.2%. 

Why is it in Orlando, FL? Florida ranks among the nation’s top 10 states for manufacturing and is home to more than 18,600 manufacturing companies. Top manufacturing segments include advanced manufacturing, aerospace and defense products, metal fabrication, and medical equipment. Florida’s manufacturing industry is highly integrated with the global economy, with Florida-origin exports to the world reaching $58.6 billion in 2014. Manufacturing employs 336,200 people in Florida, an increase of 6,700 jobs over 2014. Florida ranks #1 for aviation manufacturing attractiveness and #5 for high-tech employment in the U.S. Orlando is a top location for entrepreneurs.

Examples of types of Additive Manufacturing. 

A good overview of different types of additive manufacturing processes can be found here. Two examples from that page:

Material Extrusion (FDM)
Material Extrusion is a group of AM processes that create the final product by selectively depositing material. They do so by extruding the material through a small diameter nozzle. The base material often is a paste or a plastic. In the case of the paste a syringe type
applicator can be used to deposited the paste. For plastics usually a plastic filament is fed through a heated nozzle that melts the plastic so it can be deposited. 

Fused Deposition Modeling (FDM) is a process that uses a reel of plastic filament. Once deposited the filament will stick to underlying layers and neighboring filaments and will almost directly solidify. Due to the nature of the FDM process overhanging features should be held by support material. 

Fused Deposition Modeling showing the part (dark blue) and support material (light blue). The extrusion head (see detailed view at the upper right) remains at the same z-level and the build platform moves to allow stacking layers.

Due to the simplicity of the layout of the process and the machine, FDM is used in preferably used in the many homebuilt AM machines like for example the RepRap and the Ultimaker.

Material Jetting (Objet, Solidscape, 3D systems)

Two groups of processes are recognized that both use a jetting technology, like found in normal inkjet printers. Material Jetting is the process were the 3D printer uses an inkjet head to selectively deposit product material. 

Two types of material are predominantly used in this group of processes; wax and photopolymers. Some processes are able to directly jet metals. The advantage of this group of processes is that the nature of the process allows to change the product material during a build. In this way graded material properties are possible.The material jetting system as used by Objet. 

Real World 3D printing Success Stories
GE Fuel Nozzle - From this article

The end result is an engineering marvel, one monolithic piece that has replicated the complex interior passageways and chambers of the old nozzle down to every twist and turn thanks to the miracle of direct metal laser melting where fine alloy powder is sprayed onto a platform in a printer and then heated by a laser, and repeated 3,000 times until the part is formed. 

What makes the new nozzle so special isn't just that it has converted a many-steps engineering and manufacturing process into just one. It is also a miracle of material science since it happens to be both 25% lighter in weight, as well as a staggering five times more durable than its older sibling, all of which translates to a savings of around US $3 million per aircraft, per year for any airline flying a plane equipped with GE's next generation LEAP engine, developed by CFM International, a joint venture between GE and France's Snecma (Safran).

Why does Wenzel America go to RAPID?

While this data from NIST is out of date, look at the % of AM parts in the industries just 5 years ago. 3D printing is becoming more and more prevalent. How do you inspect these parts to make sure they are in tolerance, that they meet the design specifications?

We do 3D Scanning. We have a variety of sensors (X-Ray, Laser, White Light) that you can use to digitize your parts. From reverse engineering, non-destructive testing to metrology, Wenzel America has a possibility of scanning your part. We already partner with companies who use subtractive manufacturing. With our diverse portfolio of metrology solutions we can also provide metrology, NDT and Reverse Engineering solutions to the Additive Manufacturing world as well.

We are at RAPID to meet you and to help you answer the questions in your manufacturing process. Giles Gaskell, Wenzel America Ltd will be presenting on May 16th, 2016. (Check the Event for specific time). This workshop provides a comprehensive introduction to 3D scanning technologies, software, and processes, highlighting the differences among data capture technologies including live hands-on demonstrations of some of the most popular scanning devices. Guest speakers and demonstrators are amongst the world’s most knowledgeable and experienced practitioners in the field. Stop by Giles’s workshop and come by our Booth 455.

Come to RAPID to investigate the 3D printing and AM possibilities. 

Stop by our booth to find out how we can help you.