2020 HBK Technology Days

Dr. Peter Miller, Chief Engineer - Battery, Millbrook Proving Ground

This presentation will start by giving an overview of the various automotive battery applications (from SLI to EV) and the battery requirements for each. It will then look at test methods and standards for automotive batteries for xEV’s that cover abuse, performance and life and give some examples of these. Finally the split of testing between cells, modules and packs will be discussed.

Dr. Matt Smith, Centre for Research into Electrical Energy Storage & Applications, University of Sheffield

The presentation details the results of a series of tests to determine the Dynamic Charge Acceptance (DCA) performance of small form-factor carbon-enhanced VRLA cells designed for use in Hybrid Electric Vehicle (HEV) applications, together with standard lead-acid and lithium iron phosphate (LFP) cells. The results demonstrate how varying the conditions and parameters of the standard DCA test regime can provide a superior evaluation of DCA performance and lead to a better understanding of cell behaviour under real-world conditions. It also demonstrates the importance of recognising the limitations of existing test procedures and how they should be considered before using results from such tests to make judgements about real-world battery performance.

Mr. Umesh Tiwari, SMM, Advanced Materials, Malvern Panalytical

The electrodes in a lithium-ion battery undergo reversible electrochemical reactions as lithium enters and leaves the atomic structure of the intercalated lithium compounds. Particle size, shape and crystal structure of electrode materials play an important role in Li ion diffusion and their transportation during charge-discharge cycling. Reversibility of these electrochemical reactions with cycling largely governs the lifetime of the batteries. With certain battery chemistries, these electrochemical reactions render the electrode materials unstable state, pushing them into irreversible physical or structural changes, ultimately leading to battery degradation with cycling. Understanding the intricacies of these electrochemical reactions is, therefore, an important step to improve battery performance. X-ray diffraction (XRD) and scattering is well-suited to study these atomic phase changes, as well as a tool to understand and optimize the pathways that lithium uses to move through the electrodes. However, XRD investigation of battery materials requires special considerations that are different form the routine powder diffraction measurements.

This presentation will review the techniques related to dimensional measurements in battery electrode materials and how these relate to the battery capacity. Special focus would be on X-ray diffraction and scattering methods including considerations for experimental design. Some examples will be shared on how these considerations are applied to cathode material analysis, including Rietveld refinement, to quantify phase mixtures and atomic structure. A case study on the analysis of NCM based batteries with in-operando XRD, to track phase changes and potentially the degradation mechanisms during charge-discharge cycling, will be shared.

Dr.Ing. Prashant Khapane, Engineering Operations Manager, Jaguar Land Rover

Our products today are becoming complex, are becoming smart, our products are becoming hyperconnected. Thanks to smart, connected technology and the IoT, organizations are being buried under a flood of data from their products and processes. Digital Twin is often thrown at us as a solution to solve all the problems. This presentation will look at digital twin from automotive engineering perspective, discuss challenges involved and will ponder over the fact if we are asking the right question, does automotive engineering need a digital twin in its classical definition?

Mr. David Ewbank, Technical Director, VI-grade

Vehicle design for durability is a long-established discipline, affecting all aspects of vehicle engineering. However, despite well proven techniques and processes, the loads used for component and system design are measured on a mule or even a competitor vehicle at the beginning of a vehicle programme and may not be revisited until well into an advanced prototype stage. At this stage, any durability concerns due to the use of approximated load cases can be difficult to implement due to costly tooling updates.

This presentation demonstrates the use of advanced simulation methods to investigate the variability of a vehicle design on the wheel loads and to produce a set of load cases which are robust and encompass the loads that could be achieved considering all possible tuning variations. As well as using robot driver technology this process can be extended to introduce the human element through the use of VI-grades advanced driving simulators with a human driver to study the impact of human variability of predicted wheel forces.

Mr. Chris Polmear, Principal Engineer, Millbrook Proving Ground

The presentation explains how 5G technology is used for connected vehicle testing at Millbrook Proving Ground. It gives an overview of the onsite networks available, and an explanation of our approach to connectivity, vehicle data, and data processing. It includes examples of real-world use cases involving data acquisition from test vehicles.

Dr. Paul Gardner, University of Sheffield

In the age of big data many engineering companies are collecting large quantities of data from their systems in operation. However, much of this data represents normal, benign, operating conditions, and reveals little information about what data will look like when my system is damaged, or what will the system response be to an extreme event? In light of this problem, it is useful to be able transfer knowledge gained from one dataset or simulation to current operational data, aiding diagnosis of what the data represents. Transfer learning, a branch of machine learning, is designed for exactly these scenarios, allowing knowledge to be moved from one dataset to another. This presentation discusses exciting innovations in applying transfer learning to engineering problems, with motivating examples from non-destructive testing and aerospace applications.

Dr. Josh Hoole, University of Bristol

Probabilistic design approaches provide a route to assessing the reliability of safety-critical structures, potentially yielding more optimum designs, increased service lives or a quantification of the level of safety within a structure. However, one of the inhibiting factors that routinely prevents the implementation of probabilistic approaches is the lack of available data to statistically characterise the variability present within engineering parameters. Fortunately, the rapidly growing field of ‘big-data’ now provides greater opportunities for capturing engineering design datasets than ever before. This presentation will demonstrate a novel data source, in the form of real-time aircraft tracking, which has been exploited within the development of a probabilistic fatigue methodology for aircraft landing gear structures.

Dr. Andrew Halfpenny, Chief Technologist, HBK Prenscia

"Fatigue is the progressive weakening of a material caused by cyclic or otherwise varying loads, even though the resulting stresses are well within the static strength limits. The art of fatigue simulation is to be approximately right rather than exactly wrong." - Prof. Keith Miller

The fatigue design of mechanical systems has historically followed a 'deterministic' process. That means, for a given set of inputs they will return a consistent set of fatigue life results with no scatter. In reality the inputs are statistically uncertain -- they have an expected value and a variability. Deterministic design methods take no implicit account of uncertainty. In practice, the designer applies a safety factor to each input parameter along with an additional safety factor to the final result to allow for 'modelling errors'. In most cases, the engineer is fairly certain that the simulation results are conservative, but cannot state with any confidence what the final safety margin, reliability or failure rate will be. Furthermore, it is almost impossible to qualify the simulation by experimental testing because the test lives are significantly higher than the conservative simulations would suggest.

In comparison, a 'Probabilistic Fatigue Simulation' method is 'stochastic' in nature. In this presentation we review the concepts of 'Stochastic Design' under the following headings:

1. Uncertainty sources: including Epistemic (or reducible) uncertainties, and Aleatoric (or irreducible) uncertainties;
   
2. Uncertainty propagation: including 'Analytical methods' and 'Numerical simulation methods' covering:

• Monte Carlo Simulation;
• Design of Experiments (DOE);
• Response Surface Analysis;
• Reduced Order Modelling (ROM).

3. Uncertainty quantification: including Design for 'Reliability', and Design for 'Robustness'.

Probabilistic Fatigue Simulation offers many significant advantages over the traditional deterministic design approach:

1. Calculate life & variability in life;
   
2. Identify the most critical input uncertainties:
   Optimize for design uncertainty;
   Avoid large safety factors;
   Minimize risk exposure.
3. Design for Reliability:
   Understanding the shape of the statistical life curve;
   Design for an acceptable reliability target;
   Quantify warrantee exposure;
   Manage maintenance schedules;
   Cost vs. risk vs. benefit analysis;
   Maintain good brand reputation.
4. Design for Robustness:
   Understanding the edge cases;
   Design for the worst-case scenarios.
5. Validate simulation against physical test data:
   Validation includes mean time to failure and the statistical distribution.

A case study is presented.