By Ken Gifford
Whenever I am presenting on this topic, I have a little video I like to show. In the video, we see a man driving a cement truck. He seems happy and content with his work while he sits at the stop
light. He glances down at a photograph of his wife taped to his dashboard, and a smile appears on his face. Suddenly he gets an idea. Turning the truck around, he heads home to pay his wife a surprise visit. Upon reaching his destination however, he notices a shiny new convertible parked in the driveway with the top down. Giving it a curious glance as he walks by, he is stopped cold as he passes by the big picture window. Through it he can see his wife. Standing before her is a tall handsome and well-dressed gentleman holding a bouquet of flowers. Smiling, she gives the man a big hug. Furious, our hero makes his way back to his cement truck. Angrily he backs it up to the car in his driveway, aiming the chute from the truck into the nice leather interior. He opens the valve, dumping freshly mixed cement, filling the interior of the vehicle. Satisfied with himself, he walks back over to the picture window and peers in. Inside, he can make out the man and his wife, standing next to each other smiling. Moving into view is a photographer. The photographer takes their picture while the nicely dressed man hands his wife a set of car keys – keys for the shiny new convertible she’d won, parked in the driveway…
I love this illustration as it paints a vivid picture of what can happen when someone reacts based on the information they have – information which is incomplete. Initially it seemed like the cement truck driver reacted appropriately, based on what little information he had at the time. Unfortunately for him, a second glance revealed the whole picture. This can happen – and does quite frequently – in the manufacturing world. Often, process and quality managers are forced to make decisions on incomplete or even incorrect data, because the amount of data they actually have to work from is very small. This is due mostly because the information is collected manually via hand written logs. These are typically in the form of an operator shift report, or sampled at regular intervals during a production run, perhaps once or twice an hour. The data is then manually entered into a spreadsheet. These snapshots of information, even if accurate, do not paint the whole picture. In addition, it becomes reactive in nature, since the data doesn’t reach those who need it until much later. A valid argument could also be made in regards to the accuracy of the data. Collection challenges exist, such as operator workload causing inaccurate or missed entries. Even the number of “hand offs” in the data trail – hand written reports handed to data entry personnel, who by the way may have a hard time interpreting the handwriting of the logger, can cause incorrect data to be captured.
All of this is designed to make the case for automating data collection. Automating the collection of data removes the problems created by attempting to collect the information manually by delivering accurate and complete data to a central repository where it can be viewed and acted on. Data collected this way can provide actionable information in real time, and can even alert decision makers of a potential issue before it becomes a major problem. My goal through this series will be to demystify the process of implementing an automated data collection system, and debunk any misconceptions that may exist in regards to how difficult a system like this is to use and maintain. For example, the most common tool for users of these systems is Microsoft Excel, software that they are very familiar with and use every day. Also, manufacturing equipment today comes standard with numerous digital communication options, making it easier to network with other systems than ever before. This is the essence of the term “Internet of Things” (IoT) that is mentioned so often in trade magazines today. But that doesn’t mean that if you have older equipment you are left out in the cold either. Numerous devices exist that allow information to be collected from equipment where no such communication options exist. I will go into these topics in more depth in the next two articles. First I will go “under the hood” on data collection infrastructure. Things like communication protocols, types of network topologies, and strategies for designing a robust system. In the third and final article I will go into detail on information delivery methods and give some examples on how our customers use systems like this to improve their processes, become more efficient, and even discover “hidden gems” within their plant.
