Why quality is the obstacle to mass adoption of 3D printing

Additive manufacturing (AM), also known as 3D printing, is a hot topic.  Although the technology was invented in the 1980s, only now is it getting industry traction.  In fact, in 2017 Gartner’s famous hype cycle predicted that additive manufacturing is moving past disillusionment and into real product applications.  Many predict AM will cause massive industry disruption. Not only for manufacturers but also for everyone involved in the manufacturing supply chain. This includes transportation and logistics companies, retailers, and many others. In a 2017 study by PwC, 74% of participants agreed that companies investing in 3D printing today will have a significant competitive advantage.

Figure 1. 74% of participants in a 2017 PWC study agree that investment in 3D printing today is critical for competitive advantage. Source: PWC ‘The Future of Spare Parts is 3D, 2017.

There are many reasons for the rapid growth and interest in 3D printing for manufacturers.  For example, the actual printing process is getting faster. CAD software providers are providing feature-rich solutions explicitly for this purpose, and there are viable use cases beyond prototyping or home use.  But there is one huge obstacle to adoption in the largest, most productive use cases – quality.

A killer 3D printing use case – spare parts

Take the example of spare parts.  Most asset-intensive businesses aggressively manage spare parts inventories. This is not only because they tie up capital but also because a failure to have a spare part available in a crisis can derail operations.  3D printing is viewed by many as a game-changing solution to spare parts problems – just print what you need, just in time.  In the same PWC study, only 10% of German industrial firms surveyed use additive manufacturing for spare parts currently. However, 85% expect to do so within 5 years.

Under traditional operating models, quality is inherent in the manufacturing of spare parts.  But quality of spare parts produced by 3D printing is anything but certain. For spare part 3D printing to be considered reliable, the quality must be:

  •       Proven and repeatable under stable production processes – similar printers, materials, operators, etc.
  •       Consistent across locations and operations, under any conditions
  •       Guaranteed without input from the part’s designer

This isn’t easy for companies whose core business may have nothing to do with manufacturing.  Here are 3 risks when an asset-intensive business shifts from spare part procurement and management to manufacturing parts themselves.

Quality of source materials

Process manufacturers understand that quality in equals quality out.  Often relationships with materials providers are closely guarded or even contractually protected.  Because many 3D printing equipment and solutions vendors have little to no experience with manufacturing, the burden for sourcing materials for additive manufacturing falls on the business producing spare parts with their printers.  Unknown quality source materials present a large potential operational risk. This is especially true if the parts are used in mission-critical equipment or have a role in the production of quality-sensitive products such as medical devices, food products, and many others.

Quality in the manufacturing process

Ensuring that a specific manufacturing process produces quality parts is a science. It is a combination of advanced engineering, materials science, and flawless operational execution.  For spare parts manufacturers, providing 3D printing solutions with consistent quality standards may be a disruptive model for supply chains and solve many operational challenges.  But for spare part users, manufacturing on their own may be constrained – at least for the near future – to a fraction of their spare parts inventory that isn’t mission critical.  In a model where parts manufacturers provide a blueprint for companies to print on their own, ensuring quality in the manufacturing process is essentially impossible.  Even if 3D printing is consistently producing parts, ensuring quality requires insights from the OEM/blueprint designer. On its own, this isn’t a scalable model.

Quality control

Every traditional manufacturing process has quality control built into it, that vary from manual inspection to the application of artificial intelligence and machine learning to advanced manufacturing operations.  But since 3D printing is a constant, linear process, it tends to lack rigor with quality control (QC). Because the industry is just emerging, there are no clear solutions to QC.  One emerging idea is to apply visual inspection to the 3D printing process.  This option is fairly cheap if done at scale. It can be done at the printing site– it only relies on software and a high-definition camera. But it does require the parts designer to train the machine learning algorithms.

To learn more about how visual inspection works, check out this short video.


In the coming years, we will surely see QC solutions come to market designed specifically for additive manufacturing.

Additive manufacturing is shifting from a promising technology to a powerful disruptor, and its gaining momentum with complex ties to many adjacent transformative technology movements such as the ubiquity of the Internet of Things, the industrial use of digital twins, and the public and private investments in Industry 4.0 transformations.  IDC forecasts that in 2018, worldwide spending on 3D printing will be nearly $ 12 Billion.  Every manufacturer should think about its additive manufacturing strategy.  But at the same time, responsible large-scale adoption must be done with an eye towards quality.

The post Why quality is the obstacle to mass adoption of 3D printing appeared first on Internet of Things blog.

Internet of Things blog

How smart manufacturers bring Industry 4.0 principles to quality

Quality matters. Many manufacturers will tell you that poor quality affects both the top and bottom line. They will tell you that the consequences of poor quality are on the rise and that social media can make quality issues devastating to an OEM’s brand. They will also tell you that they are struggling to respond. The problem? While the consequences of poor quality are rising, so too are margin pressures. Margin pressures means that budgets are tight and few can respond in traditional ways. Smart manufacturers need to think differently.

Cost of quality is on the rise

Beyond brand impact, poor quality also has a real impact to the bottom line. We must scrap or rework defective products which can eat into overall equipment effectiveness (OEE) measures and lead to plant inefficiency. Challenges only increase once a defective product leaves the factory. Growing supply chain complexity means that products are increasingly costly to find and recall.

Flex – one of the world’s largest electronics makers – estimates that for every $ 1 spent in product creation, they spend $ 100 creating resolutions to quality problems.

Flex on quality for smart manufacturers


That is incredible. Solutions may include tracing root cause analysis to identify problems upstream with raw material inputs or downstream with the manufacturing process. It may also mean going back to product development to understand how product design contributes to quality.

Different industries; different impacts on quality

The implication of poor quality changes depending on the industry. For example, volume matters in the electronics industry. OEMs produce dozens of products per minute which means that a defect in a cell phone casing or circuit board assembly can be replicated hundreds of times before they identify and resolve the issue. That is a lot of potential scrap or rework. Smart electronics OEMs look for solutions that impact inspection speed. They need to quickly identify a quality issue, understand the root cause, and implement a resolution before hundreds of defective units are created.

Automotive companies have a different challenge. Here, OEMs are focused on manufacturing precision. Millimeters matter and a high degree of automation means that poorly calibrated equipment can cause small but meaningful variances. Sometimes these variances can be small – so small that only highly trained human inspectors with sophisticated testing tools can spot the difference. Smart automotive OEMs look for solutions that can help human inspectors identify these very small deviations with more accuracy.

Traditional inspection methods can be costly, slow, ineffective, and sometimes dangerous

Traditional manual inspections can be problematic. Quality inspection is a high-pressure job and the sole reliance on humans without new tools or methods can be slow and ineffective. Humans make mistakes. We get tired and have bad days. We require extensive training to spot defects and retraining to keep pace with new models. All this can hinder agility – a problem which intensified as our labor force ages and retires.

Some OEMs are getting smarter

Increasingly smart OEMs across a range of industries are approaching IBM about ways to get smarter with their quality inspection process. These firms are looking for help bringing technology – specifically machine learning and AI to bear on the problem. Fortunately, IBM has a number of solutions that can help.

Some firms are seeking to better understand the factors that contribute to quality. Have we exceeded quality thresholds? Does temperature or humidity play a role? What about equipment age and maintenance cycles? IBM has a statistical-based solution – called Prescriptive Quality – that dynamically weighs variables that might contribute to issues. This is a great solution when inspectors cannot identify quality based on an image or sound.

One of the hottest areas of interest from OEMs is how AI technology can identify visual or acoustic patterns related to quality defects. Can an image be used to identify a scratch on a cell phone casing or car paint job? Can acoustic sensors “hear” a poorly functioning dishwasher before the product is released from testing? The answers are yes and yes. IBM has two solutions – Visual Insights and Acoustic Insights – that use sophisticated AI to spot defects. What is even more impressive? These solutions can start with a small number of defective images or sounds and can learn over time to get smarter.

Does this mean we don’t need quality inspectors?

It is easy to position many of these AI-based solutions as replacing the jobs of quality inspectors. Yet this is rarely the case. Smart companies see these solutions as tools that help quality inspectors improve throughput and effectiveness. Put simply, technology like Visual Insights or Acoustic Insights help inspectors inspect products more quickly, with fewer misses, and fewer false positives. Rather than replacing inspectors, these technologies become important aids that help OEM respond better to the rising cost of quality without sacrificing margins.

Take the next step

Want to learn more about how IBM views quality and how we can help OEMs address quality challenges?

Learn more about quality management in the era of AI here.

Get more detail on specific quality solutions for your business:





The post How smart manufacturers bring Industry 4.0 principles to quality appeared first on Internet of Things blog.

Internet of Things blog

Robot swans to measure water quality in Singapore

singapore swan robots measure water quality on resevoirs

Robotic swans are being deployed in Singapore’s reservoirs to provide real-time assessments of water quality. The project is the culmination of work by the city state’s national water agency and the National University of Singapore.

Despite the best efforts of conscientious scientists, not all IoT solutions blend into their environments. Technology and utility tend to be prioritized over aesthetics. Unless you live or work near Singapore’s Marina, Punggol, Serangoon, Pandan and Kranji reservoirs, that is.

A joint project involving national water agency PUB, the National University of Singapore’s (NUS) Environmental Research Institute and the Tropical Marine Science Institute aims to gather data in a less conspicuous manner.

Read more: 100,000 IoT sensors line canal in China’s ambitious water diversion project

An elegant IoT solution

Designing a robotic swan that’s convincing to the human eye – albeit from a distance – is one thing. But the team behind the project has also fit each swan with all the tools it needs to move around reservoirs and sample water quality.

Using wireless technology, each swan is able to transmit live results to PUB, removing the need for teams to be sent out to take samples manually.

According to Channel News Asia, the SWAN project (Smart Water Assessment Network) will be used to monitor the City State’s fresh water pH, dissolved oxygen, turbidity and chlorophyll. All of these elements are used to determine the overall water quality.

Professor Mandar Chitre a member of the team behind SWAN from the National University of Singapore, said, “we started with a number of smaller bird models before we decided on the swan. It’s just the right size. If you look at it in the environment, it looks just like a swan swimming around.”

Read more: Underwater Antarctic robot Icefin prepares for Jupiter mission

Water-based robots combine with IoT once again

This is not the first time that scientists have looked to the natural world for inspiration when designing robots for use in water.

Last year, a similar project from EPFL in Switzerland developed a robotic eel to report on the water quality in Lake Geneva. Unlike the SWAN project, EPFL’s Envirobot was designed to mimic the movement of its real-life equivalent. But both have provided researchers with a way to measure water quality remotely.

With the addition of more data points and increased autonomy, it may not be long before more of these robots are spotted roaming our rivers, reservoirs and oceans.

Read more: Singapore companies settle on Sigfox for smart rodent control

The post Robot swans to measure water quality in Singapore appeared first on Internet of Business.

Internet of Business

Monitor your sleep quality with Arduino

While it can be difficult to get enough sleep, at least you can try to make it as restful as possible when you are in bed. That’s the idea behind this project by Julia Currie and Nicholas Sarkis, who developed an Arduino Nano-based sleep monitor for their final ECE 4760 project at Cornell.

The bulk of the monitoring device takes the form of a glove which measures heart rate using an IR sensor, along with movement via an accelerometer. Breathing is recorded using a conductive band wrapped around the user’s chest, which changes resistance depending on how it is stretched.

The Nano mounted to the glove collects this information, and transmits it wirelessly using an nRF24L01 chip to a PIC32 microprocessor on a base station. Data is then graphed nicely on a TFT display for further analysis.

You can read more about the project here and see the video below!

Arduino Blog

IoT-Driven Cold Chain Tracking Provides Quality Control on the Go

For some products, maintaining strict temperature controls during shipping is critical. Certain pharmaceuticals and vaccines require an unbroken chain of refrigerated environments to guarantee safety and effectiveness. Frozen foods need low temperatures to prevent bacterial growth, which can cause serious illness if consumed. Fresh produce depends on a specific conditions to maintain shelf life once it reaches market.

From production to storage to shipping, it takes multiple technologies to ensure perishable goods reach their destination without compromise. Not only must these products strictly stay within a specific temperature range, but handlers may also need to maintain other environmental parameters, such as changing weather-related humidity and pressure, equipment maintenance and anticipating rises in theft risk along the way. By tapping into sensors and the data they generate, AI-driven data insights are transforming the industry like never before, monitoring sensitive products during transit, identifying whether a shipment is at risk for damage and taking preventive or corrective action.

A picture of a truck.

AI-Driven Traceability

Beyond recognizing better delivery routes, AI can offer real-time assessments of the safety and quality of food and pharma products. By offering those real-time insights, opportunities are created to change the conditions affecting products. Making food and pharma two places where traceability is critical. That traceability extends to consumers’ interactions with cold chain products. AI can help in four categories:

  • Descriptive Analytics can help provide context to analytics, enabling a better understand of the story behind the data, to reduce false positives for a smarter system. It can greatly boost sensor fusion and the integration of combined data from multiple sensors, for example to add vibration alerts corresponding to the higher temperature that occur when a package moving away from the cold chain. If a package falls over in a truck, for example, temperatures won’t rise, preventing false positives for product losses. Examining data post-trip AI can rank the quality of service as well as ranking the service provider.
  • Descriptive Analytics can help provide context to analytics, enabling a better understand of the story behind the data, to reduce false positives for a smarter system. It can greatly boost sensor fusion and the integration of combined data from multiple sensors, for example to add vibration alerts corresponding to the higher temperature that occur when a package moving away from the cold chain. If a package falls over in a truck, for example, temperatures won’t rise, preventing false positives for product losses. Examining data post-trip AI can rank the quality of service as well as ranking the service provider.
  • Diagnostic Analytics enable automated decisions to reduce ping rates if data is running low, or to prioritize certain messages over the others. It can also suppress redundant data, reducing data bandwidth.
  • Predictive Analytics can provide theft forecasting based on data from a combination of sources including combinations of weather plus location, a rainy day plus low visibility day, or a holiday plus time of the day with location. Predictive analytics can rank quality of service in near real-time, even before a trip is over. It can also provide tremendous equipment insights offering everything from battery to sensor failure predictions.
  • Prescriptive Analytics enables maximization of good outcomes. From best mitigation factors to optimizations for the most efficient routs, prescriptive analytics streamline operational efficiency across the cold chain. This includes optimizing green strategies for reduced carbon emissions and lower energy costs.

Blockchain for Cold Chain

There’s never been a more exciting time for those working in the cold chain industry. But what does the future hold? Looking ahead five to 15 years, blockchain promises stronger layers of security for AI-driven networks. With enterprise interest in blockchain heating up, 39 percent of all companies, including 56 percent of companies with more than 20,000 employees, are already looking at blockchain implementation. Linking blockchain solutions to existing product journeys could very well provide even stronger traceability across the entire cold chain — from farm to pharmacy.

To stay informed about Intel IoT developments, subscribe to our RSS feed for email notifications of blog updates, or visit intel.com/IoTLinkedInFacebook and Twitter.