As cloud-based sensing and actuation along with compiling of data expands, we need to realise the lack of commonality in our understanding. Sometimes it’s just the words we use, sometimes it’s semantics, and still other times it’s our confusion of expected outcomes.
In machinery we talk about RPM revolutions per minute or SPM Strokes per minute or spindle speeds or IPM inches per minute, etc. all of these terms are machinery related. Parts per minute, (PPM) is what we really care about. When we look at IoT and our desired outcome, it’s the measurement and metrics where we cloud our data and confuse our information. Therefore, we need to make sure we are clear in the information, says Joseph Zulick is a writer and manager at MRO Electric and Supply.
On a manufacturing floor, you will often hear a supervisor ask if we can “speed up” the machine to achieve the new higher demands for a product or to achieve a deadline. This is where things get cloudy. The supervisor doesn’t specifically care how you get more parts, their goal is to meet the demand.
Optional solutions may be to run the machine faster, this is also ambiguous because turning up the motor speed or strokes per minute can result in a quality issue. In stamping it could be that increasing the slide velocity results in the forming tearing the material as it forms. In machining the fact that you are trying to take off more material in a single pass can cause heat, loss of control and a poor finish. On a press brake you can end up with greater spring back.
So, the question is how we can achieve the same part quality at a higher rate. Sensors now can monitor more and more. Using the Artificial Intelligence in many systems can provide several scenarios based on the past IoT feedback and anticipated results. As an example, when a part is transferred from one station to another on a press it takes a certain amount of time.
The faster you run you may be forced to operate in automatic single stroke which is a mode where the machine stops on top and now waits for the part to be moved from one station or operation to the next. Surprisingly, you can achieve greater speeds by running slower and never stopping at top. By staying in a pure continuous mode, the number of parts you can create is more than you can by running the machine faster and waiting for the automation. This is partially a focus of theory of constraints which focuses on determining the bottle neck. It can also be part of lean manufacturing and the systems that adapt these manufacturing concepts.
These theories are being incorporated into the Iot side which is monitoring and proving the best way to achieve the goal, enough parts to meet demand. Another solution found in hydraulic machines and in servo technology is stroke length limitation. Part of what determines how long it will take to make a part is the time per stroke. Simple math would dictate if something were running at 60 Strokes per minute, 1 stroke takes 1 second. So how can production be improved? In hydraulic and in Servo machines you can vary the length of the stroke.
If we look at the 1 stroke per second machine and the stroke length is 4 inches, it may be possible to remove wasted time in the stroke. All return time is wasted to some degree unless it’s when automation is occurring, and you couldn’t be producing anyway. Perhaps I can limit the stroke length to 3 inches. This time may save you ¼ of that second. This could result in producing 15 more parts in a minute!
This is the same philosophy that Link motion presses use by changing the slide velocity profile, in layman’s terms, it runs faster during the upstroke when no production is occurring.
Servo machines take advantage of the adjustment of the stroke length as well as hydraulic machines. This is true in all types of machinery. A side bonus of this is you reduce your exposure to risk and hazards since these areas are exposed longer with longer stroke machines.
IoT monitors the production numbers and proves the optimisation of machines and can serve to help us understand the information. We do need to make sure we are clear in what we ask to achieve. As shown above, asking to run faster may be the wrong solution.
Sensors can also be misapplied where they are used to sense the wrong thing. Just because you sense a part has come off a machine is no guarantee it was a good part, a part that actually was finished or packaged. More systems now allow for tracking of a good and bad part though the operation because removing a bad part can be more costly than the time to allow a sensed bad part to run through and be scrapped at the end. This is of course providing that the bad part is an anomaly and not a failure. This data can fire a diverter and allow a bad part to be rejected and ejected.
Too often we give up on data because it doesn’t change or solve the problem or what we are gathering doesn’t turn out to yield the results we need to act upon to make a difference. We feel that sensors provide solutions but a sensor without data gathering, and comparison is just a data point.
If I give you the number 6 and ask you to solve the problem, you can’t do it. You need the formula, or other data points to determine a trend, that is how you make a difference. It’s only with this information and knowledge that you can expect an improvement.
Once you have the context of this information you can build the limits that you need in order to make the change. There are some AI systems that analySe and interpret key performance indicators, but the problem is it’s just that a piece of data with no input context will give you output recommendations with no output context. As the proverb goes… Garbage In = Garbage Out.
We are improving data systems and notation is a huge part of this especially with analogue sensing. We need to know a value associated with analogue sensing and just as important is a real message not a useless code value. Visual cues are also very useful where the operator can understand and act upon the message without the need to escalate and wait on a supervisor or maintenance person to interpret a message and an action.
The destiny of manufacturing is dependent on data. We are currently moving through an intermediate step of advancement. We may treat systems like Alexa or Siri as all-knowing systems, but they really have a catalogue of keywords to run a program or skill.
The exciting part is what’s coming next which is comprehension to a lesser degree where the systems understand our faults in what we ask for and what we actually want to know. Our failures will lead to smarter systems of tomorrow which are closer than we think!
The author is Joseph Zulick is a manager at MRO Electric and Supply.