Why is a full system approach so important?

Kristian Kouumdjieff, 25/11/2019

At Romax, we talk about a “full system approach” a lot. Particularly with the move into electrification, the benefits of simulating the electric components together with the mechanical are often discussed. But long before that, we’ve been promoting a holistic simulation philosophy as the only way to truly understand system performance. Here’s why…

The complex interactions between all the components of a transmission can have a cumulative effect, which isn’t understood if the components are analysed in isolation. This starts at a basic level, between the core components: shafts, gears and bearings. Simulating complex physics and complex components is all very well and good, but getting the core system right is a crucial, but often overlooked, aspect. If the interactions between these core components are not correctly predicted, none of the results will be trustworthy.

In a Romax system, shafts are represented using flexible Timoshenko beam theory. This method, validated against test and FE, is widely considered the best solution, at least for long, thin shafts. When this is not sufficient, rotating components such as planetary carriers, differential casings, and more complex gear blanks can be converted to FE. When bearings are mounted on these shafts, they can be defined as conceptual stiffness components, selected from catalogs or defined using user-specified raceway and element dimensions. Calculating bearing stiffness is incredibly complex since they are highly non-linear and their stiffness depends on many factors – load, manufacturing or mounting clearance, internal details, varying contact angles between the rolling elements and the raceways, mounting fit, temperatures, etc. However, successfully incorporating all of these internal factors into a fast bearing stiffness calculation is only the first step.

The impacts of the bearing stiffness calculation ripple through the whole powertrain system simulation. First, it affects the shaft deflection, which causes the gears to misalign, which shifts the contact from the centre of the gear to its edge. A tilting moment is then applied to the gear, which causes the shaft to deflect, which is reacted by the bearings, again changing their load-dependent bearing stiffness. This then comes back through the system again and affects how the shafts deflect and how the gears contact. There may also be other components and loads in the system, such as external loads, bushings and thrust pads, modelled as non-linear contact elements. To add to this, the bearings are mounted inside a flexible housing, which interacts with the other components in complex ways.

It’s all one connected system, where all the components affect each other. Really the only way to get accurate system deflections and misalignments is to model the whole system, and then analyse it iteratively. Possibly the most complex part of this system in terms of simulation is the bearings. Accurate bearing simulation is thus critical – if the bearing stiffness is not captured correctly, the entire simulation model will be wrong. An important feature in Romax software is the ability to model bearing rings as flexible finite element parts. The analysis times with flexible bearings are up to 15 times faster than using full FE tools.

Here is how we achieve orders of magnitude faster analysis times: rather than use FE for the whole system, which is hugely challenging, not completely accurate, and almost impossible to use for a complete complex transmission system, we use a hybrid approach, incorporating a combination of finite element, analytical and empirical methods. Thus, we use the most computationally-efficient and most accurate method for every part, and combine them together into a fully-coupled system simulation. That means you get the best of everything – a blend of speed and accuracy, specifically optimised for powertrain simulation.

At Romax, we have always known that a full system approach is critical for understanding the interactions caused by complex shaft loads (gears) and supports (bearings). This underpins many of our software offerings and new developments, as we look for ways to ensure optimum component design and analysis within a whole system simulation context, using methods which find the best balance between speed and accuracy.

To find out more about full system approach, watch our webinar Using Romax Spin for detailed design and advanced analysis within the Romax Nexus platform.

Kristian Kouumdjieff joined Romax Technology in 2015 after graduating from the University of Nottingham with a Meng degree in Mechanical Engineering. Kristian is the Product Manager of the Durability and Structural Analysis (Romax Enduro), and Bearing Design and Analysis (Romax Spin) software products.