Data Data Everywhere, but Not a Bite to Eat

Data Data Everywhere, but Not a Bite to Eat

Joe Cichon, Vice President Manufacturing Technology at INX International Ink Co.

Joe Cichon, Vice President Manufacturing Technology at INX International Ink Co.

Chemical processing industry Lessons learned Industry 4.0 INX

International manufactures printing inks for the graphic arts industry and is unique in that we use a myriad of machines to produce batches from 10 pounds to 30,000 pounds, with an SKU count of over 500,000 unique items.

We have many process machines that are controlled by PLCs and software. Some PLC systems and control software generate tons of data, but much of that data is not stored, and subsequently, it becomes lost process data information.

We decided to start collecting and using our data on our processes many years ago. For the most part, hand collected data was recorded on paper or process orders, and finally, someone would enter it into a spreadsheet or download it from our ERP. These records are helpful when we have to troubleshoot a problem, but for the most part, most of the data was not used.

In the last few years, we decided to use the data passed through our machine control circuits and PLCs to help optimize our equipment. When we investigated options, most vendors we approached, estimated costs of about $40,000 to $80,000 just to give us an assessment. In two cases, the assessments came in between $190,000 and $400,000 at just one of our seven major plants.

Our business based ERP data helped manage and plan our business, but we needed detailed process data. We discovered early on that using SAP (our ERP) to process machine data was extremely expensive, and the cost to configure the software was estimated in the mid 6 figure range.

We had machines in our plants that controlled equipment and logged the data, but the problem was that no one in operations knew how to access the machine data. It was just consuming bits on a hard drive somewhere or just evaporating into space for the most part. The data that we collected helped engineers troubleshoot an issue, but engineers do not run our plants. We needed information that could be used by managers and operators and technicians. It needed to be accessible and easy to view and use. We did not have time for an engineer to spend hours sorting through tables of data tags to find an answer every time we have an issue. Even more important, we needed data that would help find and maintain optimum running conditions for every product we make on any machine.

"We do not have time for an engineer to spend hours sorting through tables of data tags to find an answer every time we have an issue"

We decided to start harvesting our machine data to find ways to optimize our machine processing systems. We spoke to a few integrators and software firms who indicated that they could use artificial intelligence or cloud data to optimize machine production efficiency. We discovered (circa 2013) that all of them were using machine sensors and indicators and gauges to determine when a machine was approaching failure. Looking at vibration sensors, temperature sensors, power requirements, etc., they can determine anomalies that indicated something might be wrong. This is useful information, and it is indeed worth its weight in gold. The ability to predict a machine failure before it happens can prevent many catastrophic failures that can shut down a line or a plant or cost you some very important customers. But what we really need is data for managers and operators to help ensure that our machines are running at optimum production and quality rates whenever we need them to run.

We are still on our journey, and there is much to share, but I would like to share some of our lessons learned for now. If you are aware of the possible pitfalls when you start, you can save your team a lot of money and a lot of time.

Key points that you might want to capture based on our experience.

• Start simple focus on individual line OEE data (Overall Equipment Efficiency).

• A solid infrastructure is critical to success. (check to make sure your connected manufacturing system is reliable, with compatible PLCs, HMI, and data transfer protocols and networking)

• Newer machines have IIOT options to get your machines individually, reporting up into the cloud and available on the internet or mobile devices and tablets.

• Make sure your team includes Managers and operators working with your installers to ensure that they get what they want.

• Operator training is critical and should be documented, delivered by an expert, and confirmed as well as Policed over time.

• Rely on machine data as much as you can instead of using operator scans or truth tables. (if there is sufficient power to the machine it is running)

• Data Cleansing is often overlooked and is also a critical step. Don’t use bad data, especially if operators (Humans) must enter information.

• If Operators enter important information, be sure your system validates the entry to keep bad data at bay. • Make sure the data is easy to see and access. Don’t rely on spreadsheets or query reports.

• Security is important. Be sure to use an IDMZ (industrial Demilitarized Zone). A significant risk for any company is ransomware (even if you do not feel you have confidential data to protect, Hackers can disable or damage your equipment and encrypt your data.)

• Artificial intelligence can be used to harvest performance data to help optimize production settings on equipment.

• The Data Scientist is an emerging field resource that we all will need to either have on staff or tap into as needed.

Weekly Brief

Read Also

Automation Enhances Product Capabilities

Automation Enhances Product Capabilities

Larry Saidman, Chief Technologist - R&D, Nordson
Use Continuous Improvement or it uses you

Use Continuous Improvement or it uses you

James Stewart, Head of Technology Programs, Huhtamaki
Using Automation To Improve Supply Chains

Using Automation To Improve Supply Chains

Venkatesh Ramakrishnan, Head of Maritime, Logistics and Retail