BROADCASTS.com

In my last article in the July issue of MTD magazine, I hopefully gave a useful explanation of what industry 4.0 will mean with respect to manufacturing production processes of which Metrology and its subsequent applied measurement is going to be an important ingredient. I indicated that the start date, pace of change and ultimate impact of i4.0 will differ for each company.
We can confidently predict that though the direction of travel is common, the journey may be intermittent and completion may vary by a few decades. Whatever your process, whether it\u2019s complex, relatively simple or complex when looking simple, there is one mantra that you need to live by before you start.
\u201cI often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely, in your thoughts, advanced to the stage of science, whatever the matter may be.\u201d Lord Kelvin (1824-1907)

In this article, I will follow on the focus on Metrology and its current state and its readiness to move to an industry 4 future. To do this I will focus on three key areas.

To keep my feet firmly in my own area of expertise, we will focus on Industrial dimensional measurement, not including surface texture. Dimensional measurement is by far the most widely applied form of industrial measurement with hundreds of dimensions on millions of drawings, all sample and first article checked, as a minimum, across every manufacturing process.

Dimensional measurement results in the UK are standardised to the metre and termed length measurement. Length has always been impacted by temperature and now, newer measurement technologies also utilise time and luminous intensity (light) as a factor of their measurement accuracy. So now, as Metrologists, we need to have a basic understanding of several of the S.I. Units and their underpinning science.

But back to the main thrust of this article. What is the current state and what can we expect in the future? By the way, I have no crystal ball, this is a personal assessment.

Technology Application
Current state: I said in my last article that metrology is still completing the third industrial revolution. A majority of companies when trying to keep on top of process control and quality, still commonly rely on manual measurements in the form of fixtures and first principle measurement tools and techniques. They will then in most cases have a Coordinate Measuring Machine (CMM) for first article or validation checks. This is still the most common scenario even in larger enterprises. Automation as will be required within the process; is still only applied by a small percentage.

The workhorses of coordinate measurement, the CMM are well established and common across small and large companies alike - now being mainstream for several decades. Their recent development has continued to eke out accuracy improvements. They are in the main highly developed with the sensor and the software developed and blend to tackle complex applications of geometry and tolerances. Well advanced in its technology development curve, the laws of physics are hindering any further radical development.

There has been an explosion of new technologies starting around the turn of the last century. In this last decade, they started to make inroads into the measuring equipment market share, particularly within the larger volume component arena. Specifically, this development has been in the field of Electro Magnetic Spectrum sensors. Utilising the use of visible light, lasers, X-Rays and photogrammetry techniques and often termed non-contact or optical sensors. The type of technology varies greatly, generally speaking they are not yet as accurate as a typical bridge CMM, but they are making significant and rapid progress in attaining better accuracies. Their ability to capture large amounts of point data in spilt seconds makes for fast measurement, but that does come at a cost, most of these systems are more susceptible to sensor noise (bad data returns). This is commonly offset in the systems by applying statistical analysis to the mass data to remove anomalous outlying data, smoothing and filtering. Averaging, turning point clouds into facetted surfaces, and applying dynamic geometry and pattern based algorithms. This is good for larger surfaces with their size and profile requirements, but generally less useful on smaller more accurate features where form is a critical function to be toleranced. Maybe this should be a future article in itself?

Future state: Though CMM\u2019s can with some management and allowances be added into a cell, fed by robots and eventually connected as part of a Cyber Physical System (CPS), they do suffer from a big Achilles heel. They are relatively slow, they will generally not be able to keep pace with the manufacturing process and may have to be built in parallel to a production line and have a serious impact on the \u2018Takt\u2019 time.

The new kids on the block, predominantly the optical sensors, do meet the speed required and as flexible sensors, they can often be mounted directly within the production line or at least on robot arms, as they may need to move around the object being measured. You can also consider mounting several of the same sensors to collaborate and make one overall coordinate system output. They can measure complex shapes very quickly. But, remember the accuracy and issues around accurate form; sometimes a CMM is still the best option and slower throughput is the price.

There is one other possible future solution. That is, as a company you do not buy a complete system, but only buy sensors. Maybe a hundred sensors or more, each one measuring temperature, a feature or selection of features and collaborating with other sensors to build up a complete coordinate and geometry output within the cell cyber system. Of course this may only be valid for more simplistic parts, but it does potentially meet the CPS model better than an independent software system. With off the shelf low cost sensors providing localised data that is coordinated via a cyber-system, maintenance and equipment failures could be easier to manage. But the challenges in designing and validating bespoke systems will be huge, especially the software application.

There is also a bigger question to ask, to which only market forces and an entrepreneurial approach by supply companies will answer. CPS requires connectivity and as I said in my last article, it needs a wealth of data, not only of the measured part, but of the line/cell/processes health and environmental changes; sensing all the parameters that can cause variation and issues to the resultant part and the process performance.

Complex coordinate measuring system suppliers are very good at supplying and supporting turnkey systems with their own dedicated software. The connectivity of the complex coordinate measuring systems into the cyber system and the collation of their data output with all the other sensors within that line or cell, have still yet to be tackled in the true sense of CPS. Single, simple sensors can already be incorporated, but complex coordinate measuring sensors with bespoke software is another matter. We are currently sat with no single solution, and so the market will decide. A Betamax/VHS type free market forces battle but on a much bigger scale, with huge risks and rewards for early adopters - unless someone sets/agrees a standard that drives this?

Finally, as the machine tool companies found out in the 80\u2019s, fixturing will need to be rethought. Most current fixturing is manually fed, again this will create issues as a typical robot arm does not have the senses that humans have to \u2018best\u2019 place a part in a fixture, so the fixture design has to be better thought through than currently it is. Is there a cheaper option than making a robot able to fully replicate human senses? Look up Industry 5 on the internet. Again, maybe a future article?

Skill and Competence
In the current state, we have a real problem here. Industry 4 is going to require much more planning. It will need engineers with broader and deeper skills as they will need to be proficient at planning for right first time. This is because there will be more at risk if it goes wrong. This has come at a time where the impact of 30 years of recessions and its sometimes misdirected cost cutting has eroded the less fashionable skills, particularly the underpinning of cross functional skills from both companies and the offer from education and training. Some of those skills are a shadow of what they will need to be. We have in far too many instances become expert fire fighters and forgotten how to plan for fire prevention, always the lower cost option.

Metrology has suffered more than most, as it has been lost or paid lip service to when considering most apprenticeships and degrees. Technology has made it almost impossible for colleges to keep in touch with modern equipment and both techniques and Universities, even with their relatively bigger budgets, cannot specialise in every area of manufacturing. Private training providers have easier, more fashionable and profitable streams to teach in; and equipment manufacturers ultimately want to sell their own product even when they are well intentioned and ethical. The educators also struggle to employ experienced experts as they are a rare beast.

Except for a very few examples, today\u2019s design and manufacturing engineers are taught very little about modern measurement and metrology techniques. This is something that must change if the UK wants to be a leader in smart manufacturing where Metrology will be a core engineering skill.

Data Management
Current state: We already have systems capable of collecting \xbc of a million points in a few seconds. The issue is what we can do with such rich Data, remember a small part of which is false data. Well ideally we would like to keep all the data, even the false data. When mined it could reveal patterns that improve our overall system knowledge. Hardware and storage limitations make keeping all that data problematic and expensive. So currently the equipment manufacturers apply statistics and summarise the data. This is based upon the manufacturers idea of how much of a summary is best to build a software system that will perform in its typical target market. At this time, we can collect 3 million points on a surface, check it against CAD and output some key but simple visual analysis. That core of summarised data which is still very large in file size could be stored; but if we measure 40 parts a day, after a year we would have a significant amount of data in storage and other than repeat a slightly different key visual analysis data set, it will not be used again. So, a lot of companies once they have the green tick of acceptance, delete that data as it is of no further use.

Future state: Data management is an area that is clear in terms of what is needed for the future but less clear on how it will be facilitated with regards to mass data from coordinate metrology systems. Every bit of measured data, in the raw and not already summarised, is useful for a cyber-system to use in process improvement and digital twin simulations. That means collecting and keeping huge amounts of data in a way that it can be cross referenced in a search and analysed for opportunity. This is required to make the most of the cyber system.

There are already some very good companies, some new and some that are reinventing themselves. These companies are offering production line management and controls systems that are on the march towards i4.0. These companies are working hard on the connectivity and data management side of i4.0 and linking that to the wider business management systems. The integration of single sensors as described, is not an issue as they are a plug-in tool to the cell or line designers. But the integration of complex measurement systems into CPS is not so easy without being dumbed down. Coordinate measurement systems are often not fully understood by those cell or line designers (a broader skill issue) and they have limited connectivity potential to the CPS cyber control system; and so true CPS connectivity is not readily available at this time.

The connectivity issues will be resolved as companies attempt to build CPS into their production and utilise current technology and thinking into those lines, but it may take many iterations. Simplistic measurement will go first but eventually over a decade (or two?) it will all become off the shelf for the mainstream manufacturers.

I read back my own words and feel I can never do this justice in an article of this size, there are many threads and connections of which metrology is a big part but not the only part. I have probably left you with a lot of questions. But this second part of my article was intended to stimulate thinking and get you to look at your business metrology fitness, not provide an A to B map. More talking needs to be done and if my article starts that I would be happy to hear from you with your own thoughts and comments. You can find me on LinkedIn or via the Coventry University website just type in my name.