System simulation is a proven method for frontloading the balancing of multiple performance attributes of a product. However, in the automotive industry today, a large diversity of vehicle architectures and technologies exist. This results in a huge number of variants for all subsystems. It becomes increasingly difficult to manage and analyze all possible configurations.
LMS Imagine.Lab System Synthesis provides an architecture-driven approach to tackle this challenge. This article will illustrate this with an electric vehicle application case.
Consider the example of a conventional vehicle depicted in Figure 1. Multiple variants are introduced for the subsystems: 2 vehicle platforms, 5 engines, 2 transmission types, 3 batteries and 4 tire configurations. 3 model complexities per subsystem and a total of 3 load cases are considered for multi-attribute balancing. This analysis would result in 2160 simulation models. With an architecture-driven approach, these models could be represented through 1 architecture and 48 component models capable of representing 2160 configurations to evaluate. This makes it a lot easier for configuration and maintenance of the analysis.
This demonstration example presents the use case of an electric vehicle. The goal is to evaluate different electric motor and battery variants. Each variant configuration is scored on a set of attributes: range, efficiency, performance, peak power and continuous power. To analyze these attributes, different scenario’s need to be simulated. Figure 2 depicts an overview of the electric vehicle variant evaluation.
The electric motor variants are implemented as Functional Mockup Units (FMU’s). They differ in maximum torque and efficiency, characterized by the power loss map. Three battery variants are considered with their specific technology, energy content, mass and operating voltage.
Each configuration is evaluated on different attributes. Each attribute corresponds to a specific simulation scenario . Range is evaluated during a New European Driving Cycle Driving Cycle (NEDC) whereas acceleration performance is evaluated doing a wide open throttle (WOT) maneuver. The table below lists all scenario’s and associated attributes.
With a classical simulation approach this analysis would result in 24 simulation models (4 scenario’s x 3 battery variants x 2 electric motor variants). Using architecture-driven simulation with LMS System Synthesis this can be represented as studies on 24 “model assemblies”, all managed within one project. The workflow is summarized in the video below and in the next sections of this article.
In a first step a tool-neutral architecture is defined. This architecture describes the layout of the system from a simulation standpoint. The electric vehicle architecture consists of the following subsystems:
Afterwards, the connections between the subsystems are defined. Only the relation between subsystems is specified, not the actual variables that are exchanged. The end result is the definition of the base architecture (see Figure 5).
In a second step, a template is created for each of the subsystems. The template is an interface contract specifying input and output. This input and output definition can be dragged and dropped on the template ports.
In Figure 6 the battery simulation template is depicted. The battery exchanges voltage and current with the electric motor template. Current and voltage are added, as input and output signal respectively to the port. State of charge (SOC) is added to the controller and consumed energy to the performance analyzer.
Besides the interface contract, the template also consists of exposed parameters and variables.
The exposed parameters for the battery template are: temperature, total energy and initial SOC of the battery pack. The variables are OCV, SOC, voltage, current and energy. The simulation architecture is the result of defining simulation templates for all the subsystems (Figure 7)
Defining architecture and templates will increase control and collaboration.
The template acts as a target for the subsystem designer ensuring integration in the overall system. The architecture is the framework for integrating models developed in different departments and created in different tools.
In a next step instrumented models are created. They are a combination of a behavioral model and a simulation template. The instrumentation process consists of mapping ports, parameters and variables between template and behavioral model. Figure 8 shows the port mapping of battery simulation template and the behavioral model implemented in LMS Imagine.Lab Amesim. Parameters and variables are mapped to the exposures of the template in a similar way.
All 10 subsystems are instrumented and saved in an instrumented model library (Figure 9). There are 3 battery and 2 electric motor variants. The electric motor and gearbox are implemented as Functional Mockup Units. The rest of the subsystems are modeled in LMS Amesim.
Instrumentation increases the modularity and reusability of models. They don’t need to be redeveloped, but rather can be reused in future projects.
Afterwards, a model assembly can be created. For each template an instrumented model is selected using drag and drop. This connection is “plug & play” thanks to the interface contract. The simulation template filters out the compliant instrumented models that can be selected. Figure 10 shows that the battery simulation template can be realized by one of the 3 variant instrumented models.
Subsystem models become plug and play. There is no need any more for complex integrations like cosimulations setups, importing and exporting results.
A new study is created for the first variant configuration. It consists of 4 runs or scenario’s to study different performance attributes.
The study is launched from LMS System Synthesis (Figure 11). In the background the models are composed and the heterogeneous simulations (FMI and LMS Amesim) are started in LMS Amesim.
When the simulation is complete, the results can be plotted by selecting the variables of interest. All results can be aggregated to score the first vehicle configuration on the different attributes (Figure 12). This process could be extended to manage and execute all possible scenarios and load cases.
In a final step 6 different variants are evaluated. For each of these configurations a study is created with all 4 scenarios. The synthesis of this evaluation can be depicted in a single spider graph (Figure 13). Each configuration is scored as a percentage for each performance attribute. This allows us to get a complete overview of configurations, capitalize on knowledge and improve design.
This article illustrated the value of architecture-driven simulation with LMS System Synthesis.
Models are transformed into modular and reusable assets. This allows for easy configuration, variants’ evaluation and multi-attribute balancing.