Using Smart Factory to Improve Manual Manufacturing Operations

Introduction

In the past, there have been a multitude of solutions that focus on manufacturing machine performance. Machine Monitoring solutions, I-IoT platforms, SCADA systems, and more have proliferated to help optimize machine uptime and performance. But manual operations such as assembly have been largely overlooked in these efforts.

However, there is a new set of solutions focused on video capture and artificial intelligence that address this gap. These solutions can automatically detect whether workers are following standard work, detect potential quality events, identify sources of performance variation, and much more.

In this webinar, we will review how these solutions work and how they help to dramatically improve performance of your manual manufacturing processes.

The Future of Work is Transforming

We are all faced with embracing the new reality. This new reality is faster paced, with more turnover in the workforce than ever before. It demands new approaches to manual operations and new solutions to support those practices.

Meeting these challenges is significantly easier with digital technology – such as solutions to empower frontline workers with the information they need to do their jobs – even in the face of a retiring workforce, increasing product complexity and the immediacy of customer demands. A great example is the augmented reality capability shown here in the picture. Both new and experienced workers can benefit from the highly visual work instructions to perform their tasks. This is just one of the technologies we will cover today.

Manufacturers Face Operational Challenges

What is the current state for most companies? Companies can no longer rely on traditional hard copy or video work instructions and basic training methods and expect a highly efficient workforce.? Nor can they rely on the memory and skill retention of their operators and technicians to handle the increasing complexity and variation of products and production lines. The more traditional methods of enabling front-line workers to be productive & do their jobs well cannot keep pace with the demands of today’s manufacturing environments.

Some examples:

• The retiring workforce (or those simply walking out the door) are taking their expertise with them. Companies are struggling to capture their expertise before they walk out the door.
• The use of paper-based systems takes a long time to create and even more time to keep up to date on the shop floor as the process is updated, new products are introduced, and other changes take place. Outside of work instructions, there is a lack of current documentation for compliance and safety.
• This also leads to inefficient and costly training for new or cross-trained workers. Several companies I’ve worked with have a hard time keeping newly hired employees through the training process due to the clunkiness of the experience.
• In addition to issues around work instructions and training, there is a lack of visibility into causes for quality and production issues in manual operations. Companies have been able to implement machine monitoring systems for years that automatically track faults when machines go down or produce bad parts. But this capability has not existed for manual operations until recently.

How to Provide Help

Now let’s take a look at several of the Smart Factory solutions that can help address these challenges.

As always, if you are interested in these technologies, please reach out to me at [email protected]

Video Analytics & Spatial Computing

First, let me talk about one of the coolest technologies I’ve seen in the past few years. I have a partner that works in the video analytics space. From our research, they are at the forefront of this set of products and provide unique capabilities in the space. They do for manual operations what machine monitoring does for equipment optimization. They work with the customer to install cameras above each work area, then apply AI/ML to that video to create data about each operation and each step within those operations. To me, watching this software do its thing is like magic. It can generate incredible insights about the causes of variation within each step and operation for both quality and cycle time. It can determine whether standard work was followed or if a deviation occurred. And for every deviation, for every long cycle, there is video of exactly what happened. If you want a technology that will improve root cause analysis – this is it.

Let’s walk through a few examples of what video analytics can provide:

Traceability

To get started with the software, you train it on what a good cycle should look like. This simply involves having your most trusted person perform the task below the camera multiple times. Once it is trained, the software can then detect when people are not following standard work instructions when they perform the task. This is called “automatic anomaly detection” and provides a video record of each cycle – good and bad. The video for each cycle flagged as an anomaly will be instantly available through many paths in the software interface.

• Using video search in the software, you can do Root Cause Analysis as simply as using You Tube video searching
• You can be alerted that there was an error, and seeing the video for that alert is as simple as clicking a button.
• You can quickly go down the path to determine the RCA - there is no guessing - you can see exactly what happened in seconds
• You will also be capturing video on good parts. So if a customer challenges you with a quality inquiry, you can look up a product by serial # and share the video of all operations with your customer to show evidence that you performed good work

Reducing Defects

The system can show you when an operator does not complete his work in adherence with standard work.?This data can drive how you provide ongoing reinforcement for standard work and for the training you provide to your people.

Your IE or Process Engineer can view the video and see exactly what happened, i.e., what went wrong, what steps did the operator take, and where did he not follow the expected standard work steps. Then you can use this data to collaborate and identify actions to take to improve – whether this means working with the operator to reinforce their training, or perhaps modifying the standard work to make it easier to perform.

The ergonomics analysis that can be performed with video analytics can also drive significant reductions in both the median time and the cycle variation. For example, I worked with a customer that saw the location of the tooling was causing wasted motion and significant process variation. They invested in belt holders and saw immediate benefits across all their shifts. This is one way to pitch the solution to reduce the “big brother” fear that these solutions can inspire in the work force. It may not be that the operator is lazy – it may be that the standards need further optimization.

Improve Line Efficiency

This same type of analysis can help increase line efficiency. When there is a long cycle you can immediately determine where the bottleneck station is and use video to see exactly what is happening to cause the bottleneck. You can answer questions like:?were the tools not functioning, were there enough parts in the hopper, or is training reinforcement needed

Variability on the line one thing that kills your ability to hit the desired numbers repeatedly and video analytics can show you where the outliers are. Then an IE can dig into the details without having to capture the data first - he can simply use the data that has video tied to every data point.

Your MES system may be able to give you data about what happened, but it will not tell you why. This is the most important thing available from the video streams.

Accelerate Training

It is on the job guidance and training reinforcement. The video can be used to contrast the best practices at the station with how that operator performs the task. This can be extremely powerful in guiding desired behaviors. This gets operators that are new to the task up to peak levels of performance much faster than traditional methods.

Advantages of Next-Gen MES Solutions

Another partner of mine provides what they call a “frontline operations platform” or a “composable MES solution”.?To me, the platform is an incredibly flexible solution for collecting information from the shop floor, sharing information such as work instructions or visual controls with frontline workers, and coordinating the overall flow of information on the shop floor.

It provides these capabilities with a drag-and-drop development interface along with a library of pre-built components that get you up and running very quickly. Once it is operational, you will find that there is practically no end to the use cases you can perform with it.

Let’s look at a few examples.

Within quality, you can deploy something as simple as user-friendly data entry forms for inspection or audit execution. The apps can also be significantly more complex to support inline quality tracking. Also, you can directly improve quality performance by implementing error-proofing techniques such as pick to light systems.

There are also use cases across production in areas such as receiving, packaging, and more. One of the use cases not shown here is around autonomous maintenance and providing the operator with a list of tasks to be performed that day along with work instructions on how to perform them.

Not only can the system provide support for the training process itself, but you can also use it to manage who is trained on what process, what certifications exist and who has them, and ensure that people scheduled on different jobs have all the training required.

The system also provides critical support for tracking and monitoring within the facility. This can be automated monitoring directly from machines or providing an easy interface for manual tracking.

Augmented Reality

Augmented reality is an interactive experience of a real-world environment where the objects that reside in the real world are enhanced by computer-generated perceptual information.

This is another technology with many important use cases on the shop floor.

AR solutions can deliver in-situ assembly & operator work instructions. This can feel a bit like working in the future, but there are applications in manufacturing where having hands-free and in-context instructions greatly increases operator performance and safety.

The solution can also quickly and easily capture the knowledge of experts, enable more effective training, and easier access to remote SMEs.

Improving Manual Operation Performance

Now that we’ve seen what each of these technologies can provide, let’s take a step back and look at some use cases from the point of view of operations instead of technology.

Training

As I mentioned earlier, I have a customer who is having issues with turnover on the shop floor. Not only have they had a lot of employees leave since the arrival of the pandemic, but they are having issues retaining new employees, with some even walking out during the training process. Clearly, the current on-boarding process is not capturing the hearts and minds of new employees. We talked about many things to improve the process, but here are some key items that we landed on.

Make it more fun and engaging: let’s face it, many training processes were designed with previous generations in mind. Younger workers will have more fun and be more engaged when doing training with cool technology like AR goggles and feedback from video analytics. The increased levels of attention will also help them retain that information better when they get out to the shop floor to perform the actual work. And having the video analytics in place will help provide feedback when they deviate from the standard work and continue reinforcing the correct way to do things.

Connected Work Cell

The connected work cell is central to these systems. The fundamental idea here is to make it easy to share information with the front-line worker and to collect information from the process. The collected data is then visible to anyone in the organization (that has authorization). It can also be used for visual controls on the shop floor to help guide and influence behaviors.

One basic component of the connected work cell is the display of work instructions for the current step. This should provide the ability to navigate and view all instructions, with full support for multi-page document. These steps should allow for data collection and the acknowledgement of completion of each step. This can be accomplished with traditional displays or AR interfaces for hands-free interaction.

Another basic component is data collection from the process. This can have both manual and automated components. Where there is equipment at the station that can be connected or sensors added to the process to collect data, this information can be automatically gathered. ?It can also be contextualized with the particular item or SKU currently at the station, who is working the process, etc. Then it can be stored and used for analysis at any time. Simple (or complex, if required) manual data entry forms can also be provided to allow the operator to manually input data to the system for things that cannot be automated in any way.

Standard Work

There is a critical interaction between smart factory and standard work. We’ve discussed some of the obvious touch points already and covered the delivery of standard work instructions to the operator and visual controls a minute ago. But the final bullet in this list is also critical. The capability to detect and elevate attention to deviations from the standards is key to reducing variation on the shop floor. If you think of a standard fishbone or Ishikawa diagram, a couple of the standard branches are for Man and Method. These categories (along with materials) are probably at the root of most of the variation in your process from a quality or cycle-time perspective. Being able to automatically detect deviations to the standards will help drive down that process variability and significantly improve results.

Autonomous Maintenance

I mentioned autonomous maintenance earlier, but it is worth drilling down a bit deeper here.

One of the pillars of TPM (Total Productive Maintenance) is to have the operator take as much ownership of maintaining the machine as possible. It's sort of like doing your own basic maintenance on your car - it will be more efficient and probably done with more care and attention since you own the car. But a lot of people don't change their own oil (or even wiper fluid). Other than being too busy, the biggest hurdles are some combination of: not knowing how, not having the equipment, or not having the consumable (oil, fluids, etc.).

It's the same thing for operators and their machines. They are measured on production, so they don't have the time. The equipment and consumables aren't at the machine. They don't have the instructions of what should be done when. They're unsure when to perform the tasks.

With the smart factory solutions being discussed today, many of these issues can be addressed. The system will know how many pieces have been made since the previous activities, so the operator doesn't have to track that. The system can provide a list at any time of what activities have to be performed – and step by step instructions on how to do it.

Visual Controls

A very common way of looking at manufacturing performance is with KPIs, or key performance indicators. These are often organized into categories SQDCIME – safety, quality, delivery, costs, inventory, manpower, and environmental. As I mentioned earlier, for a lot of companies these are tracked in a very manual fashion and displayed on boards with people either coloring in the sheets themselves or printing these off from Excel on a daily or weekly basis and posting them up.

They are great because they are color-coded you can immediately see if the status is green or red for any metric. And that's great! It is very visual. It is very easy to see where the problems are. But again, this can be done in a much more automated fashion with a lot less waste in the process. In the inset photo, I would argue that we are seeing a living example there of the eighth waste of the underutilization of human potential. I’m sure he could be doing something more valuable with his time than putting printed sheets into a magnetic holder every single day.

That waste is one of the key problems with manual controls. For people to walk to the boards to “enter” information or read information requires wasted movement. Writing the information on a white board and then later capturing the information in excel requires rework. And as mentioned, it is a waste of time when this could be done automatically. The poor guy in this photo is an engineering manager. He could be solving problems in the plant, but instead he is printing out forms, carrying them over to the board, then swapping out an old sheet with the new one inside of the plastic holder there.

Another issue that is extremely common is errors. Missing data can be a huge issue, as can transcription errors when someone is transferring information from the boards to excel. As an example, I had a customer that was using a pace board to track production and downtime. Then they captured that downtime from the pace board into an access database at the end of every shift. As a pilot, we implemented a smart factory system in some areas of the plant to capture downtime events. We found that the manual system was capturing less than 10% of the downtime events!

That highlights another problem – keeping up with a fast-paced environment. When it comes to capturing minor stops, high paced production counts, or any other rapidly changing information, manual methods just can’t keep up. That’s how you end up with less than 10% of events being captured.

Then there are additional limitations. Information captured on a white board is only viewable from right in front of that whiteboard. Sharing information with other groups such as engineering, maintenance, purchasing, quality or others is impossible unless they are standing right there. To save or persist the data, it has to be entered into a system anyway. I’ve worked with customers that had major product recalls that had big gaps in their available information about what happened on the shop floor when those parts were produced because the information went onto a white board but got erased at the end of that shift.

Finally, there is only so much detail that fits on a white board. A system can capture so much more context of what was happening at a machine or a process that just isn’t otherwise possible.

Quality

I’ve talked an awful lot about quality already, so I won’t get into too many details here. What I want to emphasize is that the improvements we have discussed so far will help improve all aspects of the total cost of quality.

Internally, the cost of poor quality will be reduced through things like variability detection and reduction, error-proofing techniques such as pick to light, and improving the existing quality processes with massively improved data. These improvements will help reduce the cost of good quality through reduced need for testing and inspection and the ability to replace or augment visual inspection with ML-based vision systems.

All of this helps to reduce the external cost of quality, as well. The number of quality escapes will be proportionately reduced as quality within the plant is improved. Having video evidence of exactly what happened in production is also a game changer for recalls, returns, quality investigations initiated by customers, warranty analysis and much more.

Setup Reduction

Something I have not touched on too much that can utilize the capabilities of all these systems is setup reduction. I did a full webinar and white paper on this topic, so I won’t get into too many details here. But these systems provide great benefits both during the SMED process itself, and to maintain adherence to the new process once the SMED is complete.

AR solutions can be used to easily capture the current “standard” process for analysis and document the steps. Video analytics can be used to capture the timing of the steps 24/7, not just when the expert is wearing the glasses. This helps capture all the different ways the process is done today. And the operations platform can gather contextual information from the process to further the analysis.

Once the project is complete, it is critical for people to adhere to the new process to get the benefits from the SMED. Once again, each of these systems provides benefits to driving that adherence.

Safety

Finally, there are a many areas where these solutions can impact worker safety. The one that I will highlight here is the ergonomics analysis. The video capabilities are invaluable to performing this analysis. Having this system in place will provide information for the team to analyze how the standard work and the workstation configuration impact each unique worker that have different heights, range of motion, levels of strength, etc. Easy access to outliers from a quality and cycle time perspective can show both what can go wrong in the process and how some workers may be able to consistently perform faster than standard.

Here is a longer, but not comprehensive, list:

Ergonomics Analysis

• Excess movements
• Repetitive strains

Work Instructions / Standard Work

• Startup / Shutdown
• Guidance
• Deviations
• Settings

Error Proofing

• Visual Guides / AR
• Machine Settings
• Error States

Environmental Safety

• Monitoring / Alerting
• Robot / Human Interaction

Closing Thoughts

That’s it for today. As always, please reach out to me if you have any questions or if you are interested in learning more. Each of the solutions highlighted today is from a partner for Visual Decisions. We have worked very hard to find solution providers in each of these areas that provide rich capabilities and are also good partners for our customers over the long term.

Thank you and please subscribe to the newsletter if you have not already!