054
ROBOTICS & AUTOMATION
The key to successful automation projects In any automation deployment, “The Three Ps” – Product, Process and People – are of vital importance, writes Barry Hendy of Andrew Donald Design Engineering. I have been involved in the robotics and industrial automation industry (on and off) for more than 30 years. Over the years I have seen a lot of projects make the transition from wild concept to productive and effective reality, and I have learnt a lot about what is behind both the successful and the difficult projects. That experience boils down to what I call “The Three Ps”: Product, Process and People. You must manage all three for success. It is easy to do simple automation that ‘should’ work most of the time, but life rarely travels down the middle of the road. Great and successful automation projects must take full account and manage all the things that can go wrong so they can keep going when the road gets a little rough at the edges. Achieving 99% is not good enough – that might mean a stop every five minutes. To get rock-solid automation you must take account of all the exceptions – all the things that might go wrong – because if it can go wrong … it will.
Product The first step when thinking about automation is to fully know your product. Intimately. Totally. All its nuances, variations and faults. While some customer operations are single-product, most manufacturers have a range of products they make in different batch sizes and rates. Knowing and documenting the full scope of products is the most important section in the requirements document. It is probably the fastest and the biggest that will determine the design solution, but you should also weigh that up with the percentage of your production that is spent on each product. If there is one very big or heavy product but you only run it occasionally, is it worth including in the must-have list? Just as important as the range, and for single-product operations, you must also have a deep understanding of the variation in each product. It might be seasonal, it might be temperature, it might just be tolerancing – but every product will have its variances. One project Andrew Donald Design Engineering (ADDE) did was packing croissants, but if they were left in the proofing room too long, they fluffed even more and didn’t even fit in the box! Variation in the amount a bread loaf rises is perhaps obvious, but the variance in a bottle, carton or casting might be more subtle. Cardboard packaging dimensions typically vary by about 3%, so how well a product fits in the package can change from batch of packing supplies to batch.
AMT OCT/NOV 2020
We had one project where the parts were injection-moulded and supplied to us pressed together. It is a high-rate system (about 30 parts per million) so they were both made in a multi-cavity die. While we were promised all the parts were the same, there was one combination of cavities where the parts didn’t fit together quite as snugly, so occasionally – certainly often enough to be debilitating – the parts would come apart in the descrambler. And then there is all the ways that the product can fail – or be a bit less than perfect. What are the checks that your operators might make, possibly even subconsciously, on every part? Do they know the sound or feel of how a part clips on and realise when they occasionally aren’t right? All these things can be managed, but only if they are known. If you miss them or don’t understand them, you run the risk of shipping bad product or having to add additional quality checks to a finished machine. Even documenting or quantifying what is good and bad can be a challenge. I have watched on more than one occasion while a group of the client’s team debate if a sample part is acceptable or a reject. You must know your product – completely.
Process Just as you must know your product intimately, you also have to know your processes just as well. What are all the steps you need to execute your production? And I mean ALL the steps. Yes, most will be obvious, but it is the subtle or occasional tasks that are probably just as important as quality and reliability.
What can go wrong? What doesn’t always work perfectly? What are the checks you need, the alternatives and the subtleties of all the products? Especially dig deeper for the subtleties because they are the things you might not notice, or perhaps don’t always need to be done, but can have a big impact on the finished product. I frequently see operators on a line, perhaps packing product, flip some or all the products over and quickly glance at something. When you ask the manager what they are doing they often have to ask the operator and learn that the printer might not be super-reliable, or there might sometimes be a flash from the injection moulding. Some level of Failure Mode and Effect Analysis (FMEA) is probably a good idea, but at least have a discussion with your operators on all the small things they see and do to make sure you fully understand what is needed to ensure everything goes well. Part of this will be consideration of if it is acceptable to just control the process, or if you need to check the result – somehow. Also consider that often automating a process will make it inherently more reliable since you have removed the variations natural in human tasks and brought consistency to the process, and that alone may address an issue. As your automation system is being designed, a deep understanding of your processes will be critical to ensure all the checks, fault capture and failure management is in place. Make sure you fully understand all your processes, the failure modes and the checks required.