The process of learning – acquiring knowledge through experience – is essential to the continuous improvement process that manufacturers rely on to remain competitive. Creating effective feedback loops that implement improvements based on evidence can result in huge savings in terms of scrap reduction.
These days, feedback loops can be implemented automatically in software. And the ability to draw on as broad a knowledge base as possible when doing so increases the potential for savings even further.
Manufacturers today record data throughout their processes. At Hexagon Manufacturing Intelligence, we provide software for engineering, production and metrology. Each of our software products generates a huge amount of data, whether simulating the expected performance of a part, calculating how it should be machined or measuring what has been manufactured. Sharing information between these software domains increases the knowledge base being used to make decisions, offering opportunities for new insights to improve productivity.
Let’s take an example. In a car factory, sheet metal is stamped against a solid metal die to form the required shape – a car door panel, for instance. The shape of the die is approximately the desired shape of the panel; however, after the metal has been stamped and the press lifts away from the formed part, the metal will ‘spring back’ a little, so it will not match the shape of the die. The solution is to deliberately design a die that is smaller than the desired shape, so that the stamped metal will spring back to exactly the designed shape. But how much smaller should the die be?
Simulating springback to compensate die design is a complex science, and even with the most advanced simulation techniques some trial-and-error is needed, often requiring many cycles of design and building dies – an expensive and time-consuming process. It will typically take many months to perfect the design of a single door panel.
The answer is to feed back precise measured results from the initial stamping into the spring-back simulation.
By feeding real,measured data into the simulation, it should be possible to correct the simulation and get it right second time. In time, with learning, the simulation can be improved and the ultimate goal of ‘right first time’ should be achievable, saving huge amounts of time and money.
This is just one of many examples where improvements can be made by feeding back data from one process to another – to know what you didn’t know and improve productivity. By creating a digital thread between various process stages, the full potential of intelligent manufacturing software will be unlocked. To reach the state of self-correcting systems, automated processes must work hand in hand with intelligent software to effectively and efficiently leverage the data captured by the eyes and ears of the factory.
