Bringing Predictive Analytics to the Shop Floor: OK, but How?

By Christian
Share on LinkedInTweet about this on TwitterShare on FacebookShare on Google+

Predictive Analytics can bring a lot of value to shop floor operations, especially to improve quality, yield or process sustainability, for example in composite manufacturing.

Machine learning algorithms allow to extract patterns from past production data. These patterns, which make up a model, can in turn be used to obtain predictions (“what is the risk to have a defective part?”) or even recommendations (“what can I do to reduce the risk?”).

Data Scientists wanted!

I have intentionally used the expression “machine learning algorithm” and you may think that companies that want to go in this direction need to hire a team of data scientists.

Indeed, many open source or commercial solutions require the availability of data scientist skills along with good programming skills in order to:

  • Identify the proper algorithms to use
  • Fine tune algorithms to get good results
  • Ensure scalability and performance

So it is no surprise that with the explosion of Big Data and Predictive Analytics, job postings in this field have skyrocketed since early 2012:

data_scientist_job_trends
Percentage of job offers with words “Data Scientist” or “Data Science” ©  Indeed.com.

And, as a result, salaries have soared and positions are hard to fill, which slows down the adoption of Predictive Analytics solutions.

Empowering Quality Managers and Process Experts

In order to overcome this difficulty, the DELMIA Operations Intelligence  solution for shop floor optimization (DELMIA OI) has been designed from the start for Quality Managers and Process Experts. There is no need to select or fine-tune algorithms and “correlation” is probably the most complex word used in the User Interface. Training is achieved in a few days.

shop_floor_quality
A failure analysis engineer prepares boards for corrosion testing. © Intel.

We also think that expertise is essential to obtain reliable models in the manufacturing field. For example, a process expert may identify irrelevant parameters, add relevant durations between operations, spot errors in data… And, last but not least, he may get inspiration from the model, which in the case of DELMIA OI comes in the form of human-readable rules.

Does this mean that data scientists are out of the picture? No, if you are lucky enough to have such resources, you will realize that best results are actually obtained by the collaboration between all profiles. Data scientists bring their experience on how to prepare and handle data, while quality managers and process experts can make informed decisions using their process knowledge.

The need for a Method

Even simple concepts and an intuitive user interface will not guarantee best results. You need a method to avoid pitfalls when you have to deal with potentially erroneous or incomplete data and different ways to address the problem.

Using the experience of DELMIA Operations Intelligence past projects, we have been able to build such a method, which has been recently shared in the DELMIA Enterprise Intelligence community.

The method consists in 8 steps:

understand_process
Understand process
import_curve_data
Leverage curves
 cleanup
Clean data
 prepare
Prepare data
 target
Define output
build_model
Build model
 validate
Validate model
 assess_value
Assess value

The method answers questions such as:

  • How to leverage curve data (hint: you may need BIOVIA Pipeline Pilot)?
  • Where should I put the frontier between a good and a bad yield?
  • How can I measure the reliability of the model (its ability to predict)?
  • How can I improve my model?
  • How can I evaluate the number of defective parts that could be spared if DELMIA OI recommendations were applied on the shop floor?

Discover more about how to build reliable predictive models to optimize your manufacturing operations by joining our free DELMIA Enterprise Intelligence community.

Once you are registered, it all starts with this post!

[PART 2] DELMIA Helps the Aerospace Industry Meet the Challenges of Composite Manufacturing

By Christian
Share on LinkedInTweet about this on TwitterShare on FacebookShare on Google+

Composite wing flap at Airbus
A composite wing flap in Filton, Bristol.
Source: Department for Business, Innovation & Skills, UK.

I’m Christian Chaplais, Senior Manager of R&D DELMIA Enterprise Intelligence Applications. Welcome to the  second blog  of a two-part series on how  Operational Intelligence is helping the Aerospace & Defense Industry.

Finding a Way Around the Complexity of Composite Manufacturing

Many composite parts manufacturers have been exposed to quality issues for years. Some have used classic approaches such as simple statistics, advanced statistics or optimization consulting services to find an answer, but came up short.

So, the question remains. How do you solve composite manufacturing issues without going through all the complexities? There is a way to discover and apply an empirical (data-based) model without the complications in just a few weeks: Operations Intelligence (O/I).

With (O/I) Process Rules Discovery, for example, quality engineers or process and product experts can discover patterns (or rules) explaining whether results have been satisfactory or not. This can be done with a limited number of observations, which keeps down costs, in a ramp-up study.

Process and product experts can also understand the model with Process Rules Discovery, change it by editing the rules and immediately see the impact on the rule KPIs (Key Performance Indicators) based on facts (data).

Here’s a sample rule discovered by Process Rules Discovery:

sample rule

The rule can be interpreted as:

When the product is in the autoclave for an extended period of time (cure cycle time is high)…
…and the binding strength of the fiber is low,
…and fibers have been aging sufficiently,
then the quality is good.

Let’s take another O/I example. With Operations Advisor, shop floor workers can assess risk and take preventive or corrective action in real-time. Operations Advisor recommends values for actionable parameters (settings) without requiring any change to the process specifications or investment in new material.

operations advisor

[Operations Advisor risk assessment and proposed settings ranges (in green)]

Adopting Operations Intelligence

The DELMIA Operations Intelligence solution for Composites has been widely adopted by the Aerospace & Defense Industry from both OEMs and tier-one suppliers.

For several years, one company has been faced with an important and repetitive nonconformance issue (delamination) on the composite leading edge of wings for an aircraft manufacturer. On this family of products, the reject rate could reach 13% and the rework rate 28%. There were delays (up to 6 months of manufacturing backlog), extra internal costs, a loss of confidence from the customer and internal frustration. Multiple quality tasks including process audits, investigating new processes, SPC analysis, inspections of raw material, etc. did not solve the problem.

They then decided to use Operations Intelligence to analyze two years of production. In less than six weeks, two influent parameters, unsuspected until now, were identified (the fluidity of the resin and the time during which the part is kept under vacuum), as well as the recommended lower and higher limits for these parameters. By applying the rules discovered, they managed to instantly reduce the scrap rate to zero and the rework rate to 1%, removing any backlog shortly after.

I’d like to hear your experiences with Composite Manufacturing? What was the outcome?

Continue the technical conversation. Join the DELMIA Enterprise Intelligence Community: https://swym.3ds.com/#community:453

 

A Sprinkling of Smart Dust

By Catherine
Share on LinkedInTweet about this on TwitterShare on FacebookShare on Google+

Written by Catherine Bolgar

6-18-2015 12-53-22 PM

The “Internet of Things” allows industrial companies to tune or monitor equipment and processes with sensors. But how can these be best positioned and remain connected? One solution is to use smart dust, deployed in oil refineries, industrial process automation, breweries, waste-water treatment and elsewhere.

Smart dust was conceived in the mid 1990s by Kristofer S.J. Pister, now professor of electrical engineering and computer sciences at the University of California, Berkeley. He wanted to  create tiny, diffuse wireless sensors that could be used for product tracking, temperature monitors, defense surveillance and other such applications.

By 2001, Dr. Pister’s team had developed an autonomous sensor the size of a grain of rice, containing five solar power supplies. “It was completely useless for anybody trying to do anything practical,” he says.

But with growing industry interest in smart dust, he founded a company, Dust Networks (acquired later by Linear Technology), to work on possible industrial applications.

When we got out of the university setting, we learned that size wasn’t particularly important, but power and reliability are,” Dr. Pister says.

The sensors had to be so efficient that their batteries would be able to last a decade, and communications technology had to function in harsh industrial conditions.

“We found that as we made these things better, the package gets smaller,” Dr. Pister says. The Dust sensors are now about the size of a sugar cube—too big to inhale, as some had feared—and are fixed in precise spots, so  “they aren’t thrown willy-nilly in the breeze,” he notes. “They aren’t going to wander.”

The sensing work itself—detection and measurement of temperature, flow rates, vibrations etc.—doesn’t require much power, but communicating it does. As a result, Dr. Pister’s team focuses on extremely low-power radios, turning them on only at the right time (down to the microsecond) either on a schedule or when the sensors have information to report.

To further save energy and improve reliability, the sensors are wirelessly connected in a mesh for efficient batch networking, communicating among one another to send data in short hops that don’t require much energy.

“The n6-18-2015 12-59-54 PMetworking aspect is similar to what goes on in the Internet,” Dr. Pister says. “It’s [about] how to build a reliable network when individual paths may be unreliable. We had to come up with different optimum solutions than what people are used to in radio.”

Typically, a sensor system has many low-power transmitters and one really good receiver, in what is called a “star-connected network.” However, in an industrial setting, the physics of a star network doesn’t work well. “The nature of radio frequency propagation and thermal noise in receivers mean there is some tradeoff,” Dr. Pister says. “If anything changes in the environment, it interferes.”

Sound travels like ripples when a stone is dropped into a pond, he says. When a ripple hits a wall, it reflects back and makes patterns. “If you’re a radio, you don’t want those patterns. You get destructive interference,” Dr. Pister explains. Industrial settings are full of such radio-reflecting walls.

That’s where the mesh really shines,” Dr. Pister says. You don’t need any one particular length between devices. Everybody can talk to everybody as long as [a] path is available. Packets [of data] keep going through, no matter what happens in the environment.”

The technology continues to improve and new applications continue to arise. “It’s not out of the question that 10 years from now there’s a whole new deployment,” he says. “The interesting thing, especially in the industrial process space, is that standards tend to last a long time.”

The first Dust sensors, deployed in 2007, are still working with their original batteries. While new generations may use less power, they operate with the same protocol, so companies don’t have to replace older models when a new one comes out. Linear Technology continues to sell products launched 30 years ago.

The quest to reduce size and energy use continues. One research focus is “energy scavenging,” whereby sensors can be powered by vibrations from the equipment they’re monitoring, or by differences in ambient temperature. “It’s not science fiction,” Dr. Pister says. “We have customers who integrated our wireless sensors that are running off solar power or temperature differences in a refinery. It’s part of the future of infinite-life products.”

Catherine Bolgar is a former managing editor of The Wall Street Journal Europe. For more from Catherine Bolgar, contributors from the Economist Intelligence Unit along with industry experts, join the Future Realities discussion.

Photos courtesy of iStock



Page 1 of 912345...Last »