This is the first in a series introducing the best approach, preparation and practice when system modelling. This post specifically investigates what you should consider before you start modelling.
Modelling a system can be a difficult task to get right. There are many things to consider before approaching a system modelling software that can save you a lot of trouble and make your results more accurate. Choice of component detail and target outputs are just as important as component accuracy when building a system.
In this post I’ll outline some basic principals to system modelling that could save you a lot of trouble in the long run. I’ll be using a vehicle model as a basis and for examples as a system with a range of complex, multi-domain sub-systems.
What is System Modelling?
The obvious answer is that it’s modelling of a system of components/sub-systems. But the key to it all is the efficient, accurate modelling of a system so that it is representative of the true system. It is also imperative to understand the interaction between each system. This is especially relevant when troubleshooting.
One of the key qualities of a good system model is not that of the greatest detail or highest accuracy, but applicability to investigation. Choosing the correct fidelity/detail of the system/sub-systems to suit the goal of the investigation in the most efficient way is arguably just as important.
Before Building Models
System Targets
First thing to consider is what you want/need the system model to tell you. It seems like an obvious step but is one of the most important, so is worth exploring. “What is it that you want to get out of the system model” and “what is the focus of your investigation?” These are the key drivers of both what is required in the system and the necessary detail in the components.
For example, if you’re considering the kinematics of non-compliant suspension then there is no need to consider anything like the engine or transmission. But it’s also worth noting that the masses of the components aren’t important, as shown in the video below. The kinematic motion of an ideal suspension is not dependent on the force of the system, just the movement. However, if you were to introduce compliance/flex to the system then the mass and inertia of suspension components become important; as the force balance is dependent on mass of components.
The key question is how detailed does the system need to be to achieve the desired fidelity of results?
Availability of Input Data
When building components, or subsystems, a prescribed amount of detail is required to achieve the desired accuracy. In most cases the fidelity/complexity corresponds to the amount of data needed. There is no substitute for good quality required data. If you don’t have, and can’t get, the required data for the model then you cannot count on that element of the model to be representative of the physical entity.
It is imperative to know the amount of data you need or have available before you go to make/assemble a model at the fidelity level you need.
For example, consider modelling an engine to know the amount of fuel usage on a drive cycle at a preliminary stage of vehicle development. The engine model would be located within a vehicle and coupled to a driveline that has a fidelity to match the required output. You may only need a torque curve, BSFC map, mass and 1D rotational inertia to roughly gauge the amount of fuel usage. Now consider modelling an engine for emissions. The amount of data is much larger, from camshaft lobe and inertia data, crankshaft offsets, mass and inertia data to fuel and flame quality.
This unfortunately means that if you don’t have sufficient data then you may not be able to achieve the desired detail. This can result in you not being able to achieve the accuracy that you need. This means you either need to gain more data or change the focus so that the available data is sufficient for the investigation. There is no reason in wasting time and effort building a model that is not representative or reliable.
The only exception to this is performing very time consuming validation against measured data and sensitivity studies. But even this can only validate the model for tests that are similar to the validation tests. To make this feasible, you can only be missing a very small proportion of non-critical data to make this a viable option.
Systems Are Only as Good as the Worst Model
While considering a system as a whole you need to know what the weakest component is. If you have an interdependent system then only when each component is working correctly does the system work well as a whole. This includes the level of accuracy and fidelity of each component.
Consider the case of a driveline model to evaluate the high frequencies at the output. You can have a mean effective pressure engine, a transmission model with gear mesh models but a clutch model without clutch springs; then, this could affect the interaction between the engine and the transmission.
This means that considering the previous point, it is only worth building a model when each component is of the required level of detail.
Written by: David Briant – Project Engineer
Please get in touch if you have any questions or have got a topic in mind that you would like us to write about. You can submit your questions / topics via: Tech Blog Questions / Topic Suggestion.