Digital Development in Three Dimensions

Afaster and less expensive track to new power units – this is what IAV’s “Trinity” project promises. The greater use of simulation, better intermeshing of CAD and CAE as well as the smart analysis of large data volumes can reduce the time taken to develop the powertrain’s assemblies by up to 15 percent. This can cut costs by as much as 20 percent. The new approach is suitable for developing combustion engines, hybrid powertrains, e-motors and transmissions.

Today, power unit development involves four phases. The concept stage represents the first time the idea is put into practice and needs to show on the test bench how close the developers have come to meeting the specification targets. For the initial development stage, the new unit is made from prototype tools and tested. This is followed by the second phase in which the power unit is manufactured with volume production tools. The volume production development stage shortly before SOP has the purpose of validating the development. As today’s approach uses a large number of realworld prototypes, testing accounts for up to 75 percent of costs, whereas simulation makes up approx. 10 to 15 percent.

This process can be speeded up significantly. IAV wants to reduce development time by 15 percent and cut costs by up to 20 percent. Above all, this major saving is made possible by informative simulation results which, in future, are to provide the basis for releasing developments on a large scale. As IAV precedes along the three dimensions pf physics, geometry and data in this virtual development approach, the project is named “Trinity”.

The most evident change in the “Trinity” process is the omission of the first stage. “Instead, however, greater effort is then needed in the concept phase”, explains Dr. Michael Berg,cross-sectoral process owner for digitalization in the Powertrain Systems Development division at IAV. “This will show a far greater level of maturity in future which will provide us with a very good basis for assessing basic functionality and initial tendencies in respect of durability.” However, far more detailed information on functionality and durability are then to be provided by simulation. The precondition is that developers must be familiar with the physical effect mechanisms, reproduce them in mathematical models, validate these models and, on this basis, produce reliable prognoses. The aim in future is to obtain a large part of releases for volume production on the basis of simulations.

In the meantime, higher computing capacity and better mathematical models have made this possible. “Today, for example, we can generate detailed models of the oil circuit with all components”“, Berg reports. “And we are constantly working on improving and validating our computation and simulation methods with a view to obtaining results with a high level of confidence.“ Based on this confidence level, IAV’s developers define which testing and validation activities are still needed. In other words: they scrutinize all previous test procedures and programs and modify them as necessary.

This not only reduces the number of tests but also the type of tests. In future, this will increasingly provide the basis for better understanding effect mechanisms and validating models. In addition, they can also help to improve test methods (e.g. online measurements of oil emission) or establishing new ones (e.g. wear measurement on the basis of radionuclide technology). Needless to say, they will also be used for providing information that still cannot be obtained with simulations. “This is an ongoing process because every development comes with new findings”, Berg says.

In developing the geometry, CAD and CAE tools are to be interlinked more closely in future. The aim is to estimate the impact of design changes on functionality and durability directly in the simulation model. However, this is an area in which it is still necessary to accept compromise: “High-precision simulations are extremely complex and very costly, and they demand CAE specialists as well as high-performance computing clusters”, Berg says. One possible solution consists in linking simplified simulation models with CAD data. Although they are less accurate, they save a large number of iteration steps later on with complex simulation methods.”

This phase is validated with digital mock-ups (DMU) which can be used for conducting various investigations: installation analyses (do the components fit together geometrically?), movement analyses (what happens, for example, during load and speed alterations?) and ease of assembly (which installation forces are necessary, are the clearances sufficient for volume production?). And because all data is available in digital form, the developers will be able to use virtual and augmented reality for making information more transparent.

Of great importance to the new approach is an end-to-end data-management regime that covers the entire value creation process while also making it easier to incorporate changes. The individual phases currently use different systems, making data transfer complicated and susceptible to error. “Besides automating data transfer, the big challenge lies in managing the complexity resulting from changes and the large number of vehicle models”, Berg explains.

When layout and validation take place on the basis of physical and geometric models, IAV’s developers speak of the “virtual twin” – i.e. it is model-driven. The “digital twin” includes a further dimension: real-world data from development, production and usage. This makes it data-driven. Among other sources, the data comes from test bench investigations, fleet testing, production, repair shop reports and customer feedback. “It is about capturing data along the entire value chain, combining it, processing it for use and then making it available for resolving problems or for new development activities”, Berg explains. “So, the task is to generate new knowledge from data.” Besides the knowledge IAV’s experts have, methods of artificial intelligence play a key part, e.g. association and cluster analyses, machine categorization, pattern recognition and the automatic detection of anomalies.

The three focal dimensions of physics, geometry and data are in constant interaction. “It will be important for us to mesh our work with all disciplines involved in the vehicle development process, i.e. with the development of control units, their algorithms as well as calibration”, Berg emphasizes. “Only then will we be able to tap the maximum potential in terms of cutting costs and saving time.”

A core team of staff from various disciplines – including design, computation, simulation, testing and data management – is advancing the “Trinity” project at IAV. It uses the scrum method with weekly sprints. “The specialist work is done by the experts from the respective disciplines”, Berg says. “They help us to enhance the methodology.” Company-wide involvement that pays its dividends: In future, “Trinity” will cover the development of all powertrain components and systems.