The Value of Certainty in Manufacturing
Why deal with uncertainty?...
Whether bringing a new product from conception into production or operating complex plant and production processes, success rests on careful management and control of risk in the face of many interacting uncertainties.
Today’s fiercely competitive market and increasingly stringent regulatory environment is such that there is very little margin for error. Failure to understand and manage risks can result in severe financial penalties and even damage to reputation.
Historically, chief engineers and project managers have estimated and managed risk using mostly human judgment founded upon years of experience and heritage. However, in the modern era of High Value Manufacturing, the design and engineering of products rely increasingly on computer modelling – “The Era of Virtual Design and Engineering”. When computational simulations are used, these various risks and uncertainties must be accounted for.
This ‘Era of Virtual Design’ opens the opportunity to deal with uncertainty in a systematic, formal way. The benefits of this are varied, and include;
- better management of risk attaching to key decisions,
- convergence on designs which are robust in the face of uncertainty.
The level of interest and investment in research and industrial uptake is growing significantly across the globe. To progress, the mathematic sciences must work with the statistics community and the engineering community. This was the basic premise for the creation of a Special Interest Group in the subject of Uncertainty Quantification and Management in High Value Manufacturing.
What is Uncertainty Quantification and Management?
Figure 1. Modeling and simulation incorporating uncertainty
The white boxes in Figure 1 show the route a modeling simulation process takes. This process starts with input parameters, through to the formation of a model and then to the computation of an output quantity of interest.
Uncertainty quantification adds extra detail to this process. The yellow boxes demonstrate where uncertainty creeps into the process, and can have a huge effect on the interpretation of the result results of a simulation. As such, it is vital to quantify them and incorporate them into whatever simulation is being run (a full description of the diagram can be found at http://engweb.swan.ac.uk/~adhikaris/stochastic.html)
In short, uncertainty quantification means quantifying all uncertainties;
Uncertainty in model inputs
Uncertainty in propagation
Uncertainty in model structure and solution
Exploring the Opportunities…
The UQ&M SIG hosted a two-day workshop in late June 2015. The workshop was attended by industry representatives from the manufacturing world (food & drink, aerospace, automotive, construction, defence, power, oil & gas and many others) who discussed their approaches with world leading academics. It was fascinating to see what uncertainty means to different sectors, and the varying rates of adoption.
A key aim of the SIG is to assemble a clutch of industry cases at various levels of maturity, identify gaps in industrial research and support industries overcome the barriers to adoption. The first step was to formulate problems, and the presenters on the day didn’t disappoint with the variety of problems to address;
Ride optimisation considering vehicle mass property variation (Automotive)
The setting of wing target loads (Aerospace - aerodynamics)
Uncertainty in the built environment (Built environment)
Climb-cruise engine matching (Aerospace -structures)
Contact and friction analysis of a turbine blade and disk (Aerospace – propulsion)
Dealing with model uncertainty and variability in formulated products (Pharmaceuticals)
Uncertainty and measurement (Metrology)
The influence of max. metal manufacturing practices on design performance (Aerospace – manufacturing)
Uncertainty beneath our feet (Oil & Gas)
UQ and wind turbines (Off-shore renewables)
Integrity measurements (Manufacturing)
Slides and agenda can be found here
Important themes to emerge:
Communication and presentation of uncertainty – How can these concepts be communicated to decision makers? Interactive data visualisation techniques commonplace in some sectors could provide a step change in how uncertainties are presented and, more importantly, acted on.
Input uncertainties – For many people, UQ is about propagating uncertainty – but where are these input uncertainties coming from? More often than not, it is a judgment based on all available evidence and experience. How can these be gained then? Through the process of elicitation – this is a difficult process to do well.
Subjective experience as a target for complex designs – In consumer industries success rests on not only performance, but perception of performance – but how can this subjective target be built into complex models?
Dealing with many, varied uncertainties – Well established techniques exist to propagate uncertainties through models, and they are highly effective. However the complexity of dealing with many uncertainties and many varying types of uncertainty are often prohibitive in their computational expense.
Engineering of material properties – The opportunities with new materials, composites for example, which can have their properties ‘engineered’ opens up new paradigm in design. However, new paradigms also introduce uncertainty as there is a lack of historical knowledge in how these materials behave in complex designs. How can these materials be incorporated and their uncertainties quantified?
Inverse methods across complex designs – Being able to identify which of your input uncertainties affects strongly your quantity of interest is a vital capability in the design process. If you can identify important uncertainties to focus on, measurement strategies can be designed to improve your knowledge of the quantity of interest.
The SIG will continue to bring this work to industry, by formulating problems, creating opportunities to network and addressing barriers to adoption.
With inputs taken from talks given by Prof. Tony Hutton, Airbus UK, Prof. Tony O’Hagan, Emeritus Professor of Statistics and Prof. Sondipon Adhikari, University of Swansea
Agenda and Slides
** Note: Some slides will be released pending authorisation - please check back regularly **
Affiliations in Attendance
Aircraft Research Association
Atomic Weapons Establishment (TBC)
Bourton Group LLP
EON Exploration and Production
Flamingo Engineering Ltd.
Jaguar Land Rover
The Welding Institute Ltd.
Imperial College London
Institute for Risk and Uncertainty
London School of Economics
University of Bedforshire
University of Cambridge
University of Exeter
University of Liverpool
University of Oxford
University of Sheffield
University of Warwick
Knowledge Transfer Network
National Physical Laboratory
The Manufacturing Technology Centre