Richard Leach, Patrick Bointon, Andrew Dickins, Lewis Newton, Adam Thompson and Luke Todhunter, manufacturing metrology team, University of Nottingham
The use of metrology during a process is becoming a key part of advanced manufacturing, especially due to the need for traceable and reliable data as manufacturing becomes digitised. This article discusses a 10-year roadmap developed by the UK’s University of Nottingham and High Value Manufacturing (HVM) Catapult to benchmark and stimulate R&D in the field of integrated metrology for advanced manufacturing. This roadmap is the final output of a project that involved several information gathering exercises, including interviews with key staff at HVM Catapult premises, an international web-based survey and two conferences.
The HVM Catapult supports the UK’s academia and manufacturing industry in adopting and applying technologies so as to enhance competitiveness and drive UK prosperity. Although the main area of focus is manufacturing processes, the HVM Catapult must be able to measure everything in order to understand the effect that process parameters have on the end-product, making metrology an essential skill in its armoury. As such, the HVM Catapult is in a unique position to transition the research in integrated metrology from academia out into the manufacturing industry. The roadmap guides the UK’s future investment in integrated metrology capability, paving the way for the development of a systematic approach to integrated measurement and control for advanced manufacturing systems, to identify and guide opportunities for commercialisation of early research, and to define necessary future research themes.
The following definitions and their meanings were used during this exercise to categorise the various terms used within integrated metrology (integrated refers to measurements during manufacturing, for example, during a cutting or laser additive process). Figure 1 shows the typical positioning of a defined type of measurement with respect to the manufacturing process, and figure 2 shows the relationship between the various integrated metrology terms. In-process metrology is any measurement taken that can be used to predict or detect issues during the manufacturing process. In-process measurements must be synchronised with the manufacturing process so that the process can be monitored. All in-process measurements must be integrated into the standard production process for the part being manufactured, so that they are time sensitive. If a measurement is being performed during the manufacturing process but is not being utilised for process monitoring and control, then that measurement is not in-process and is instead defined under off-line.
1 of 2
Figure 1
Flow diagram of a standard manufacturing process, visually demonstrating where each type of integrated measurement takes place.
2 of 2
Figure 2
A demonstration of how the different metrological terms relate to each other in most circumstances. These terms can have some overlap, for example, both in-line and off-line measurements may be situated on-machine in some cases.
Off-line metrology is any measurement that is not in-process. Off-line measurements are not synchronised with the standard manufacturing process and, therefore, cannot be used for process monitoring. A component measured off-line can be returned later to the process line, but this measurement does not cause a time delay in the process of other components and is not an essential part of the production line. Off-line measurements will often be conducted outside of the manufacturing environment, either on a measurement station in the factory that is separate from the standard process line (sometimes referred to as at-line) or in a laboratory. Off-line measurements can also be, in some cases, on-machine, but never in-process, due to there not being an intent for process control with the data.
In-line metrology is a form of in-process measurement that occurs right before, right after or between manufacturing stations. In-line measurements are taken on separate measurement systems along the standard production line where manufacturing is not occurring. These measurements are often taken off-machine, but in some situations may be on-machine if a single machine consists of several manufacturing stations. In-line measurements are regular and must be taken before the next stage of the process can continue, either for each individual part before it moves to the following manufacturing station or as a sample from a batch. An in-line measurement must be part of the product’s standard manufacturing process and be used for process monitoring and control.
On-machine metrology is a form of measurement that is used to record data on a station during manufacturing. On-machine measurements can determine any variable that provides information about the process. In many scenarios, on-machine measurements are in-process but can be off-line if the data is not being utilised for process monitoring and control. In contrast, off-machine would describe any measurement that is not taken on the manufacturing station.
In-situ metrology is a form of on-machine measurement that primarily records data directly from the location where the manufacturing phenomenon is occurring. In-situ measurements are used to monitor how the equipment is interacting with the workpiece. This can be through measuring of the workpiece at the point of interaction directly or shortly after the phenomenon has passed and before the station’s task is complete.
Roadmap methodology
To develop the roadmap, a number of tasks were undertaken, the results of which are summarised in this paper. These tasks were as follows.
Data-gathering at HVM Catapult sites
The aim of this task was to systematically gather the key data for a roadmap in integrated metrology across the HVM Catapult. Existing post-process metrology roadmapping data formed the starting block for this task. Each HVM Catapult was visited to carry out an extensive inventory of existing integrated metrology and to capture future requirements, as well as to understand barriers to those requirements and to identify key future R&D programmes. Planned interviews with key staff members were conducted using pre-determined questions. The visits all took place in 2019.
Online survey
To understand the landscape of integrated metrology in the UK advanced manufacturing industry, a survey was conducted to capture the current state-of-the-art and industrial requirements for integrated metrology to support and enhance UK manufacturing. The survey was made available online from November 2018 to August 2019.
Final roadmap and recommendations
A final roadmap has been produced, including an implementation plan with recommendations for the next phase of commercialisation and R&D in academia, the HVM Catapult and UK industry. This roadmap identifies opportunities and gaps for both commercial exploitation and further research in integrated metrology.
Key survey results
The online survey was created to obtain key information from the greater manufacturing industry on the current state of integrated metrology. As shown in figure 3, there have been twenty-six responses from five countries—including the UK, US and China—and 81 percent of responses have come from small-to-medium-sized business. This has afforded key insights into the current capabilities of smaller manufacturing enterprises.
Figure 3
Company size of survey participants.
Figure 4 shows the integrated metrology techniques used by survey participants. Contact probe/stylus techniques prove the most popular, being used by 85 percent of participants. Optical techniques follow close behind at 77 percent adoption, suggesting that most participants use a combination of optical and contact techniques. Techniques that focus on build environment, such as thermal and humidity sensors, are less popular.
Figure 4
Integrated metrology techniques used by survey participants.
Figure 5 shows the forms of integrated metrology used by survey participants. A high level of integrated metrology is evident, as 65 percent of participants are performing some form of on-machine measurement. However, only 54 percent of participants use in-process measurements, suggesting that some of those on-machine measurements are performed off-line and not utilised for process monitoring or control. In-situ measurements are often the most challenging to perform, depending on the manufacturing process, yet they are performed by 38 percent of participants. This is a substantial response count, given that it is a difficult measurement category, and highlights the importance of measuring tool-part interactions as a method of process control in industry.
Figure 5
Forms of integrated metrology used by survey participants.
Figure 6 shows survey participants’ plans for automating metrology techniques in the next ten years. An impressive 77 percent of participants are looking to incorporate some, or an additional, form of automation into their current integrated metrology practices, and 35 percent are expecting to fully automate all measurement systems. These significant numbers show the importance of artificial intelligence (AI) and automation for the future of integrated metrology and the need for substantial R&D in these areas.
Figure 6
Survey participants’ plans for automating metrology techniques in the next ten years.
Figure 7 shows the current limiting factors preventing survey participants from incorporating integrated metrology techniques into their current manufacturing processes. As expected, the most common barrier to adoption is cost of installation, identified by 58 percent of participants as a limiting factor. Perhaps more surprisingly, the second most common is lack of information and awareness. This result confirms the need for communication and education over the next ten years to ensure all members of industry are made aware of the available advancements in integrated metrology.
Figure 7
Current limiting factors preventing survey participants from using existing integrated metrology techniques.
Figure 8 shows survey participants’ priorities for integrated metrology over the next 10 years. The responses to this section are relatively evenly spread. The most popular is reducing inspection time, selected by 27 percent of participants. Most forms of integrated metrology, besides in-situ measurements, require a pause in the manufacturing process in order to obtain the measurement. This has an associated cost penalty, as it can reduce the total number of manufacturing processes that can be performed in a day, and this delay is exaggerated for off-line and off-machine measurements. In second place, process optimisation and component verification are equally popular, and only one vote behind them is defect detection/process control. All three of these categories call for an increase in the reliability and quality of the results obtained from integrated metrology systems. By improving both the accuracy and the density of information that can be obtained from integrated measurements and sensors, components can be better verified, and processes can be better understood and therefore better controlled.
Figure 8
Survey participants’ priorities for integrated metrology over the next 10 years.
Proposed activities to achieve the future aims of integrated metrology
This section discusses the key research and development activities that must be performed over the course of the next ten years in order to achieve integrated metrology goals.
Process integration
The end-goal for the next 10 years of process integration is to have moved a far greater number of integrated metrology techniques from off-line to in-process. To achieve this, research into the appropriate available measurement technologies must be performed for different manufacturing processes to identify what needs to be measured and which methods are most suitable. From here, the environment that the measuring system will be subjected to must be addressed, for example, by investigating temperature compensation methods for measurement systems in forging or machining environments. Novel methods of on-machine inspection may need to be developed, such as in-process optical measurement techniques that are suitable for placement within additive manufacturing (AM) build chambers.
Once the appropriate inspection method has been decided on, physical integration of the equipment into the manufacturing process must be undertaken, involving, for example, embedding sensors within the machine using a secure, reliable and repeatable procedure, or facilitating the streamlined movement of a part into a location suitable for inspection. At this point, techniques can be developed that incorporate the newly integrated sensors and measurement systems into the manufacturing process and the resulting captured data into process control operations.
Data analysis and management
The aim in data analysis is to be able to attain real-time data capture and processing for multiple sensors and measurement devices simultaneously. To achieve this, research must be undertaken to develop a workflow that collates all relevant sensor and measurement data into one central processing unit. This way, all analyses can be carried out via one system, thus streamlining the analysis process, allowing for ubiquitous use of digital twinning and simplifying decision-making challenges.
Moreover, standardised data formats must be developed that are compatible across many sensor types. This will enable multiple measurement types to be collated together in one system easily. Developments in the efficient transfer of large quantities of real-time measurement data, including appropriate compression techniques, are also required to maximise the amount of unique information that can be transferred at a given data rate. Intelligent selection of critical sensor data is also required to maximise the value of stored information and minimise the cost of long-term data storage.
In addition to the need to obtain good data, any metrology solution embedded in a factory requires: (1) data confidence and (2) data trust. As the use of AI and deep learning start to remove the need for process improvement and optimisation in the manufacturing environment, the data being used to make decisions needs to provide a high level of certainty in order to prevent the cost of quality significantly increasing due to the use of integrated metrology and, ultimately, the inability to certify the process in a high-value manufacturing environment. This will demand a new suite of codes, regulations and standards that need to be internationally realised to enable integrated metrology across the end-to-end supply chain.
Sensor development
A primary area for research in the advancement of sensor technology is the ability of sensors to withstand harsh environments in some manufacturing processes. Robust sensors that can operate at very high temperatures need development and testing so that they can be integrated into machines and facilitate a new range of in-situ measurements.
The accuracy and speed of sensors can be improved through the development and use of the latest advancements in material and electrical technologies. Furthermore, reducing the size of sensor and measurement devices will make their integration into areas of limited space, such as build chambers, possible.
Automation and adaptive control
Automation is one of the major areas for the development of integrated metrology, and the goal in the next ten years is to create fully-automated, decision-making systems. To achieve this, complete inter-machine communication is first required to enable the transfer of data from manufacturing machine to control unit and vice versa. This aspect of R&D is linked to the integration of a central processing unit and advances in streamlined dataflows.
Next, machine learning techniques must be developed and utilised to train a computer system to identify critical manufacturing changes. These include defect detection as well as sensor/part relationships to facilitate process control. The training will need to be performed in tandem with a computer system expert to ensure that it is done efficiently.
On completion of the training, decision-making algorithms can be developed that combine the trained system and real-time information from the sensors integrated into the manufacturing process. From here, control can be given to the computer system to adjust the parameters of the manufacturing process accordingly.
Design for metrology/inspection
To achieve meaningful development in the future of advancing manufacturing, it is critical that integrated metrology is intelligently considered and accounted for throughout the manufacturing process. Research needs to be undertaken to identify the most valuable and appropriate sensors and measurement techniques for each manufacturing environment and process. Using this knowledge, bespoke considerations can be made to ensure that the use of integrated metrology in the manufacturing process is much easier. This includes designing custom spaces inside the manufacturing area to accommodate measurement devices and sensors. Ease of data output should also be factored into manufacturing equipment design, so that real-time analysis can be achieved at separate, or even remote, computers and centralised processing units.
Thoughtful consideration of future inspection processes should also be incorporated into part fixturing; for example, by allowing parts to be easily removed from machines and set up at a measurement station in a controlled manner, measurement techniques can be performed easily and repeatably, improving measurement quality and value.
Traceability, verification and validation
The integration of sensors and automated controls has limited value without proper sensor calibration techniques and established measurement traceability. Considerations during the design of manufacturing tools and equipment need to be applied to allow for the inclusion of calibration artefacts and techniques for any measurement devices and sensors that are anticipated to be included during the manufacturing process. A library of digital twin models should be developed that serve as references for SMEs to develop their own validation methods from an established starting point. Standardised, agreed-upon definitions should also be produced for integrated metrology specific applications, such as defects, to better characterise the manufacturing process in a universally acceptable way, streamlining the development of automated process control and defect detection operations through machine learning.
Aside from the traceability issues specific to integrated metrology solutions, there is still a significant unsolved metrology challenge: optical instruments that produce point cloud data do not have a standardisation infrastructure and uncertainty statements are almost non-existent. There needs to be a significant push in this area. Another area that is not being adequately addressed is the quantification of uncertainty when machine learning models are applied.
Ten-year roadmap for the future of integrated metrology in advanced manufacturing
Process integration
Sensor development
Design for metrology/inspection
Data analysis and management
Automation and adaptive control
Traceability, verification and validation
Acknowledgements
The MetMap project was funded by the Engineering and Physical Sciences Research Council (EPSRC) and the HVM Catapult as a Researcher-in-Residence project (grant number: EP/R513507/1). The authors would like to thank all members of the HVM Catapult Metrology Forum (now called the Assurance Forum), Professor Mike Hinton, Dr Sam Turner and Professor Ken Young (HVM Catapult) and Dr Mark Summers (National Physical Laboratory (NPL)) for their comments and suggestions.
The complete Integrated metrology 10-year roadmap for advanced manufacturing document is available here: https://bit.ly/3b8dT84
University of Nottingham