Paul Benning Chief Technologist 3D Printing at HP Predicts 3D Printing Developments in 2019

Paul Benning is the 3D Print and Microfluidics Chief Technologist at HP. Before that he was an HP Fellow and the Chief Technologist of their imaging and printing division. He’s a noted expert in nanotech, microfluidics, and inkjet with a Ph.D. in Material Science. I’m really impressed with the caliber of people at HP and the amount of thought that they put into their technology and all of the aspects that surround their technology. I, therefore, jumped at the chance to interview Paul about some key trends for 2019. I’m glad that we got to learn that HP is able to print circuits on its machines and that they’re going to be incorporating machine learning into manufacturing. I’m also glad that Paul’s predictions are firmly rooted in practice science and manufacturing and not in “dream a little dream” blah blah like so many predictions.

Machine learning could significantly reduce scrap rates in 3D printing, is this something you would expect to happen in the near term?

I expect machine learning to escalate innovation in the manufacturing industry in the coming year. As machine learning is integrated more into 3D printers and control systems around the world, engineers and designers will be able to receive information about the temperature of the machine, what the powder looks like, binding agents used, image data and final part geometries. All of this information will aid in reducing scrap rates and ensure the parts produced are concise and fully functional.

How do you see machine learning and 3D printing interact? 

In addition to being able to share more information than ever before with engineers and designers, the integration of machine learning and 3D printing delivers the capability to monitor a part in the field. People can follow the finished part into the world and see how it performs over its lifespan, tying the findings back to design configurations. Designers can uniquely peg every party and track them via a serial number, enabling real-time supply chain traceability.

Can machine learning be used to reduce part shrinkage for example? 

Yes, machine learning will be used to improve the process during development and to provide real-time feedback during part printing and this level of control will help to optimize all performance vectors including part shrink.  An even more significant impact of machine learning might be in more precisely predicting part shrink so that the design and process can pre-compensate for the predicted deformation producing a printed part with tighter tolerances. 

Do you really think that generative design will make designers faster? How? 

Absolutely. Generative design has the potential to make designers 10-100x faster by leveraging algorithms to discover every possible iteration of a solution. Engineers can create and simulate thousands of designs – including those that they’re unable to envision themselves – in a fraction of the time. The beauty of generative design is that engineers are no longer limited by their own imagination but can instead leverage artificial intelligence to co-create better products in a faster and more sustainable fashion.

To what extent will simulation improve generative design so that parts can be optimized?

By building in simulation and testing to the design process, engineers avoid expensive manufacturing re-works. This helps optimize parts by ensuring technology is an active participant in the design process, rather than simply reflecting actual finished designs.

Do you also expect toolpath optimization for particular geometries?

For Multi Jet Fusion and Metal Jet the “toolpath” is replaced by the printed pattern and decisions of where drops of each agent are placed.  I do expect that the drop placement patterns will be optimized for particular geometries to produce precise and mechanically strong parts.

And if we do how do we feed this information back into generative design software? 

Using information shared back from machine learning monitoring 3D printed parts in the field, designers and engineers can input real parameters into generative design software, such as product size or geometric dimensions, operating conditions, target weight, materials, manufacturing methods and CPU. The software then generates all the feasible designs and runs a performance analysis for each to determine the best ones for prototyping.

I’ve always found it rather tantalizing that with the HP 3D printing technology you could put lots of coatings on objects. With porosity and surface quality being so problematic for us, is coating parts something that you’re looking at? 

Yes, we are investigating coatings that are applied both as a post-print process and during print in multi-agent systems.  This voxel control capability will allow HP 3D printers to go beyond simple coatings to produce patterned surfaces – different coatings in different locations – and even enable materials control away from the surface, inside the part.

Can we expect circuits and conductive materials from HP 3D printing? 

We have demonstrated printed circuits at HP Labs and shown operational sensors like strain gauges.

With binder jetting metals won’t we always have a problem with part shrinkage? And won’t that always be terrible because shrinkage will vary due to part geometry and size? 

In any technology where we start with powders and create dense final parts there will be shrinkage as the open space between powder particles is consumed.  We can produce some quite complex geometries today and I expect that as we continue to learn our models will improve and we will provide design tools that work with predictable shrinkage and clever support strategies to provide the broadest design space.

How can you ensure that these parts have the right tolerances?

Well-characterized production machines that are operated using good process control methodologies will give high confidence in producing parts with the expected tolerances.  Advanced computational modeling and machine learning will help us automate the design process to get the best design quickly and will help automate the production tool setup and control processes so we build the best part every time.

What are some application areas for 3D printing that you see opening up?

We’ll see accelerated impact of digital manufacturing take hold in the form of production applications, particularly across the automotive, industrial and medical sectors. In the auto sector, we’ve seen an increased focus on developing production-grade materials for auto applications as 3D printing gravitates from prototyping to full production of final parts and products. Additionally, as new platforms such as electric vehicles enter mass production, HP Metal Jet is expected to be leveraged for applications such as the light weighting of fully safety-certified metal parts. Industrial 3D manufacturing also enables the automotive industry to produce applications in new ways that were previously impossible, along with the ability to design application-specific parts for individual systems or models.

LLNL machine learning detects 3D printer failures with 10ms of video

Engineers and scientists at Lawrence Livermore National Laboratory (LLNL), California, have applied an algorithm to detect flaws in parts as they are 3D printed. Convolutional neural networks (CNNs), used with real-time in situ monitoring cameras present the next step toward an ability to “fix it on the fly” and improve the reliability of metal additive systems. “This […]

3D Systems Looks to Increase 3D Printer Efficiency with Aquant’s AI Platform

For the last two years, 3D printing industry giant 3D Systems has been looking into product launches in hardware, materials, and workflow, in order to create more opportunities for additional applications. 3D Systems President and CEO Vyomesh Joshi (VJ) mentioned this strategy again at RAPID + TCT in Texas last month, and highlighted a few examples.

For instance, the company has been focused on applications in the medical and dental fields for a while, and both the US Air Force and the US Navy will be utilizing its technology to reproduce older plane components and qualify metal 3D printing for warships. 3D Systems also recently introduced a new metal 3D printing system, along with an integrated metal 3D printing software platform.

Never one to just sit back and rest on its laurels, 3D Systems rallied after disappointing Q3 17 financial results and outlined a fairly aggressive approach to keeping its market leadership position. In keeping with the plan, the company recently made its newest announcement – it’s chosen the Artificial Intelligence (AI) platform by New York-based Aquant to increase field service efficiency through parts prediction.

“Advanced technology is key to continued growth for our business. By applying Aquant’s AI technology to our service processes, we believe we are taking a major step towards the vision of providing our customers with zero unplanned downtime,” said Mark Hessinger, the Vice President of Services for 3D Systems.

Aquant, an enterprise AI platform, learns the unique language of other enterprises through machine learning, and uses this knowledge to increase equipment uptime – what it refers to as Uptime as a Service. Its machine learning can provide a step-by-step troubleshooting process, which allows its customers to make faster, smarter decisions, driven by hard data, by taking Aquant’s recommendations for “predictive actionable service.”

The company uses Natural Language Processing (NLP) algorithms to quickly convert both historical unstructured and structured data into a helpful knowledge base. Aquant’s predictive AI can help elevate organizations by increasing first-time fixes on machines, which completely negates unplanned downtime.

Shahar Chen, Aquant’s CEO and Co-Founder, said, “We are honoured 3D Systems, one of the leading 3D printing companies in the world, has decided to partner with us. Our technology will allow 3D Systems to leverage all of the data they’ve gathered over the years to create an immediate improvement in speed and accuracy of issue diagnosis, achieve a significant increase in machine uptime, reduce operational costs and provide fast ROI.”

3D Systems will maximize its 3D printers’ uptime through Aquant’s innovative AI platform, which will allow the company to diagnose machine failures more accurately and quickly. Thanks to its increased productivity, 3D Systems will be able to save money by cutting out repeat service visits. In addition, Aquant analyzes historical item usage, so it’s better able to forecast any future demand.

According to the Aquant website, “Even the best experts cannot predict the exact parts and skills necessary to complete each job. In order to maximize machine uptime and increase first-time fix rate, Aquant’s machine learning algorithms predict which parts and skills are required for the job.”

The technicians at 3D Systems can call on Aquant’s technology to quickly diagnose 3D printer issues based on their report symptoms. In addition, the company will be able to better predict which parts will need service calls, escalate complex problems to the next level without delay. – basically letting Aquant take care of all of the heavy lifting.

It’s smart decisions like this – teaming up with Aquant to reduce 3D printer downtime – that keeps 3D Systems on top.

Discuss this and other 3D printing topics at 3DPrintBoard.com or share your thoughts in the Facebook comments below.