Distribution operations need data more than ever before. But what should they be doing with it? Eric Rice, senior offering management lead with Honeywell Intelligrated, has answers.
SCB: We hear a lot about the need to make data-driven decisions, but what exactly are the data points that need to be analyzed in order to make that happen?
Rice: You have to look holistically at the warehouse. You want to see what’s happening on the asset side, but there are also key performance indicators inside the control system. And when you get up to the warehouse execution layer, you're looking at how you're optimizing labor.
SCB: With all of the data and systems that are available today, is there a danger of being inundated by too much of it, and not being able to distinguish between what's important and what’s not?
Rice: There is a danger of that. I've seen customers gather tons and tons of data, but then they never look at it. It's important to pick a small scope where you can find some value, then start to grow from there. That's how I would recommend you start on that transformation.
SCB: Maybe part of the answer lies in connected technology with the industrial internet of things. How might that help to solve this problem?
Rice: When I think of the internet of things, I think of a cloud platform taking data from a control system. When a customer implements controls today, there are usually what we call islands of automation. Maybe you have visibility into one piece, but you have to go to another system to see the data in another part. We want to combine all of those islands of data, so you can start to get a picture of how the entire warehouse is performing.
SCB: Is combining that data really that easy?
Rice: It depends. With modern technology, there are a lot of open platforms. An easy one to leverage is OPC [Open Platform Communications]. Most controls vendors support some kind of OPC interface. You can start there, expose data on the interface and start collecting it across systems to get a better picture of the entire warehouse.
SCB: The technology that provides you with that data can be extremely varied. You've got Bluetooth, RFID, barcode, cellular. How do you consolidate the data coming from so many different communications platforms? Is that an issue?
Rice: I don't think it's an issue in today's world. There are a lot of different devices you can use to transpose data from Bluetooth to serial or Ethernet or whatever kinds of protocols you need to get that done. The bigger challenge is what format the data is in. Take a simple example, like a read rate. It’s located on a particular type of control system, and maybe in a different location in the memory map. The challenge is how you combine all that data together to create the same level of visibility in all parts of the facility.
SCB: The initial purpose of getting all this data is to understand the state of things as they are right now. At the same time, companies want to be able to take that data and apply it in a historical fashion to improve their future operations. Is it possible today for the data to serve two purposes?
Rice: I think it's possible, but we have to be realistic about how reliable we can be predicatively. I think today, since we don't have all the data, we haven't figured out all the scenarios to train a machine-learning model. We train it on what we have at our fingertips, maybe for the last three months, then we look to see where it deviates. It’s more anomaly detection than a truly predictive outcome in today's world.
SCB: When you refer to machine learning, are you also talking about artificial intelligence?
Rice: People use those two interchangeably a lot of times. I don't feel like it's artificial intelligence. If you're talking about processing images and making better decisions based on that processing, that's what I would call machine learning.
SCB: Sell me on the benefits of doing this whole exercise in the first place. Why should companies be so concerned with this idea of monitoring the data and creating a connected technology environment?
Rice: There's a ton of downtime that happens throughout the day in facilities. There are throughput goals that aren’t being met, and inefficiencies in the process because they don't have visibility. They’re leaving money on the table at the end of the day. When you consider how much it costs to handle a high no-read rate in a system that's running over 200 cartons a minute, that can be over a hundred thousand dollars a year just by someone having to touch those boxes and put them back on the system, or printing a label for them. If you can eliminate some of your air cases, lower your recirculation rates and increase your first-pass yield as well as have a predictive-maintenance model in place, you can probably eliminate 40% of your unplanned downtime.
SCB: How do you see the future? What types of new capabilities might these systems acquire in the years ahead?
Rice: There's going to be more image processing. Instead of having a bunch of individual sensors for different areas, you might have one camera that can look at a larger area, see how cartons are flowing, and make better predictions around that. With 5G, you're going to have a lot more connectivity, with more and more devices linked together. That's giving you a bigger pipe. That will alleviate some of the limitations we have with bandwidth.
Timely, incisive articles delivered directly to your inbox.