IIoT project success: First identify the problem, then begin collecting data

March 11, 2020
In this installment of Automation Zone, learn how to sidestep the rookie mistake of collecting data before you know what you’re looking for.

Sam Hoff is president and CEO of Patti Engineering, a trusted advisor providing automation system design, hardware, software, installation, and Industry 4.0 services for various clients, as well as a member of the Control System Integrators Association. For this interview, Plant Services editor in chief Thomas Wilk spoke with Hoff about the growing need for maintenance teams to engage with non-traditional partners like IT and system integrators in order to implement proactive maintenance programs.

Automation Zone

This article is part of our monthly Automation Zone column. Read more from our monthly Automation Zone series.

PS: When maintenance and reliability teams start looking at integration projects, how do you recommend they begin?

SH: You start first by properly defining what your biggest problem is. What could I figure out on my plant floor that I can take care of using data? For instance, I was at a client site recently and one of their biggest problems is work-in-process waste. They had poor communication from the back-end of their process, their suppliers, to the front-end of their process, their scheduling. We’re implementing some cool RFID solutions around helping them track their work-in-process inventory better and reduce waste.

So, the first thing to do is define what’s important to you. Is it OEE? Is it machine life performance? Is it predictive maintenance? What problem are you trying to solve? Everything has to start with a problem!

Once you figure out what problem you need to solve, the second thing is to determine how you can measure that problem. You have to figure out how you get the data off the machines or systems. What sensors (if any) do you add to the system? What kind of architecture are you going to use to get that data?

I think one of the common mistakes I see is that many people want to start by collecting tons of data and assume the data will create an epiphany. While data collection is very important, I think you’ve really got to start with a problem, and then work on how you’re going to collect the data from there.

PS: When you meet with clients, what types of support do you find they look to you for first as their integrator partner?

SH: A lot of it is technology selection. It always works best when the client comes to us and says, “We have this problem, and we need to solve it. What do you think are the best practices around solving this? What kind of solutions do you recommend? What kind of edge devices?”

I have been into a lot of facilities, and they’re collecting tons of data, and there are tons of visual displays, but no one is doing any analysis on the data. I hear from managers, “I have 80 data parameters that I’m collecting on every part that I built, but between keeping the system running, managing employees, and meetings, I have no time to analyze it.” You must think about how you are going to analyze the data that you’re collecting.

A lot of times, once you’ve defined the problem, you’re going to start collecting the data. You have to massage that data and understand exactly what that data is telling you, and that’s part of the analytics aspect of it. It is going to be an iterative process.

We’ve heard from several practitioners that they were surprised that data cleansing took so long. We’ve heard that story at least three times: “We embarked on a digital project, an IoT project, and we figured we’d spend most of our time in the analysis phase. And in fact, it was the data cleansing phase where we spent most of our time.”

There’s a lot of data out there, but it’s a lot of dirty data. A lot of these systems were never built to produce data, so a lot of the data has to be verified to see if it’s truly accurate. It’s interesting, we see data analytics being used a lot in health care and banking, and from sites like Facebook; well, that data is a lot cleaner than data that you’re going to get out of manufacturing, there’s a limited number of fields, and people are filling out the fields. In industry, you have to ask yourself different questions on the data that you’re collecting. “How often do you need to collect the data? Do you need to collect it every 10 milliseconds? Do you need to collect it every 10 seconds?” You can spend a lot of money on data storage by collecting data way too fast for what you really truly need.

PS: With projects that center on asset management and asset/machine health, do clients come to you with known issues? Or is it the case where a client figured there were some new technologies that could help them measure asset health, and they approach you before they’ve got specific problems defined?

SH: A lot of what we hear, especially around plant maintenance, has to do with predictive maintenance. For instance, “warn me a month beforehand that this pump motor is going to fail.” For that motor, there are all kinds of condition monitoring systems that you can put in to help measure that. We are also seeing gains in tool wear. A tool manufacturer may say a tool is good for 20,000 cycles, but much like an automotive tire, the life of that tool will depend a lot on how you use it. It may last only 10,000 cycles or it could last 50,000 cycles depending on use.

PS: We get a lot of questions too about brownfield assets and equipment versus greenfield. How do you help people account for older or obsolete equipment?

SH: With brownfield sites, you’ve got to take a look at each individual asset and what kind of data you can get out of that asset instead of just adding sensors and a separate device to try to collect that data.

One of the things that I see that happens a lot of time on brownfield assets is that, instead of going into the asset and trying to get the data directly off that asset, somebody will come in and add sensors outside of it and then put on some type of edge device. Now you’ve got duplication of sensors, or duplication of equipment out on the field, and the maintenance department now has more sensors and devices to manage, and you’ve actually slowed them down.

These newly added sensors are not critical to the operation of the machine, therefore they become a low priority. I’ve been in companies before that maybe have 50 inline inspection systems and 30 of them are bypassed. At a management level, they do not even realize it because they are not monitoring these devices and they are not critical to their production. They are critical to their quality process and they are not capturing the defective parts they should be.

Part of your brownfield strategy also has to be how you’re going to process the data on the edge as opposed to the cloud. There are all kinds of conversations around the edge and what you can do, what the devices look like, that sort of thing. There are some cool technologies out there. However, you don’t want to add a whole bunch of industrial computers to your process that are just more devices for you or IT to maintain. The better question is, if you install 50 edge devices, how can you manage all those devices from one single point and have them self-manage themselves to a degree.

When it comes to digital conversations, especially in regards to digitalizing asset management and asset health, do you talk to IT first? Have you gone out to the operations and maintenance side first, to make those contacts?

We’re traditionally an OT company, so usually it’s operations coming to us. But you’ve quickly got to get IT involved and on board. I see the OT integrator business model becoming more like an IT integrator business model which is more life-cycle based, more recurring revenue, more asset management, and less just project and T&E based.

PS: What’s your sense of the quickest way to bring people on board who are encountering this side of their business for the first time?

SH: One of the things I have noticed is that teams have got to be very much collaborative between different departments in the facility. For instance, when you’re starting to get data out of these systems on the floor, the controls department, the industrial engineering department, and the maintenance department need to work closely together. There’s a lot more people that have to come together to help institute these connections and understand what the data is telling you.

The silos within an individual manufacturing facility really need to start breaking down, and the most successful projects that we’ve had in the past are where those silos are breaking down.

Sponsored Recommendations

Reduce engineering time by 50%

March 28, 2024
Learn how smart value chain applications are made possible by moving from manually-intensive CAD-based drafting packages to modern CAE software.

Filter Monitoring with Rittal's Blue e Air Conditioner

March 28, 2024
Steve Sullivan, Training Supervisor for Rittal North America, provides an overview of the filter monitoring capabilities of the Blue e line of industrial air conditioners.

Limitations of MERV Ratings for Dust Collector Filters

Feb. 23, 2024
It can be complicated and confusing to select the safest and most efficient dust collector filters for your facility. For the HVAC industry, MERV ratings are king. But MERV ratings...

The Importance of Air-To-Cloth Ratio when Selecting Dust Collector Filters

Feb. 23, 2024
Selecting the right filter cartridges for your application can be complicated. There are a lot of things to evaluate and consider...like air-to-cloth ratio. When your filters ...