The vision is clear: Model-Based Systems Engineering (MBSE) is now the cornerstone of a fully digital engineering ecosystem. Organizations like INCOSE champion this approach, and the Department of Defense (DoD), in its digital engineering strategy, lays out critical imperatives aimed at dramatically improving the efficiency of developing and fielding complex advanced cyber-physical systems.
Think of unmanned aerial vehicles navigating intricate airspace, sophisticated spacecraft exploring the cosmos, or the next generation of agile and adaptable 6th-generation fighter aircraft. The ultimate goals are stark and necessary: reduce the time and cost associated with delivering new and upgraded systems.
Let's delve into the key imperatives driving this transformation and explore how modern MBSE is the linchpin in achieving them.
Today's engineering landscape is dominated by an unprecedented surge in system complexity. Cyber-physical systems are no longer isolated entities. They are intricate networks of interconnected hardware, software, sensors, actuators, and human interfaces. The rise of autonomy, where systems make real-time decisions with minimal human intervention, and the proliferation of "system of systems" architectures, where multiple independent systems collaborate to achieve a common goal, exponentially amplify this complexity. Managing the complex relationships, dependencies, and emergent behaviors within these systems, using traditional methods, is becoming increasingly unsustainable.
For decades, the engineering of complex systems has been largely driven by a document-based approach. Requirements were captured in lengthy text documents, designs were communicated through static diagrams often created in disparate tools, and analysis was performed in isolated disciplinary silos. This "status quo" suffers from critical limitations:
Modern MBSE offers a powerful paradigm shift, providing the capabilities needed to overcome these limitations and accelerate the delivery of complex cyber-physical systems:
MBSE leverages formal modeling languages like SysML, and more specifically, the recently adopted SysML V2, to specify requirements with precision. Such representation enables computational analysis for consistency, completeness, and specification correctness. MBSE models precisely specify various system aspects such as system components hierarchy, system interfaces, system states, system functions, functional decompositions, and system constraints. Capturing all those aspects is nicely supported by SysML and other MBSE languages. In addition, MBSE tools support integration with requirements tools, such that textual stakeholder requirements can be easily traced to model-based specifications.
For example, SodiusWillert ReqExchanger synchronizes requirements from tools such as IBM DOORS into IBM Rhapsody.
Formal models can be validated by execution against simulated scenarios, acting as a digital twin of the designed system. This involves complementary techniques, each validating certain aspects of the specification.
➡️ Functional Simulation
In a functional simulation, the model is validated against scenarios by injecting stimuli processed by the model's behaviors. This can be done at various levels:
The goal is to show that the model produces expected outputs and meets a set of constraints. It checks interfaces, input processing behaviors, and completeness relative to the scenarios. It is called “functional” as non-functional aspects (e.g., durations) are typically ignored.
MBSE tools like IBM Rhapsody support this, though simulation technologies vary.
This calculates and optimizes parameters described in the model, such as minimizing fuel consumption. It typically requires solvers, ranging from basic calculators to optimization engines (e.g., linear programming). IBM Rhapsody can use solvers like Maxima or the MathWorks toolbox.
It combines system models with physical prototypes or emulated environments. It involves mathematical models and tools like MATLAB-Simulink or Modelica to simulate environments (e.g., fluid flow, control laws). IBM Rhapsody interoperates with Simulink for hosting plant models.
Model-based testing uses a model to describe test cases. These are used to test the model and may later be adapted for the system implementation. Test cases can be derived from system models to improve coverage and reduce manual test design. Models can also simulate test environments for early testing.
IBM Rhapsody includes an MBT module ("Rhapsody Test Conductor") that uses scenario descriptions (e.g., Sequence Diagrams) to validate requirements. These test cases can also be reused on implementations, such as those based on AUTOSAR.
MBSE's approach facilitates the evaluation of various design alternatives against key performance indicators (KPIs) within the model. Automated trade studies and optimization algorithms can help identify the most effective solutions based on defined criteria. IBM Rhapsody supports optimization of system parameters by hosting various solvers such as Maxima.
Modern MBSE tools transform system-level models into inputs for disciplinary engineering tools, ensuring consistency and design accuracy. This “downstream” integration replaces the traditional document-based handoff, where specification documents are passed to discipline engineers. Document handoffs are error-prone and less effective. Achieving a “digital continuity” from systems engineering to discipline engineering is one of the hallmarks of the "digital threads" architecture.
MBSE models can generate:
The Digital Thread, a continuous flow of data throughout the system lifecycle, is part of modern MBSE. Model transformation specifications are central to the digital handoff process, translating key aspects of the MBSE model (typically the logical design) into an architecture defined by a target implementation technology.
Model transformation technology plays a crucial role in this, as it should be easy for domain experts to specify the mapping from a logical system model into, for example, a software component architecture.
M2M for IBM Rhapsody by SodiusWillert is a model transformation framework that allows table-based mappings of System components into implementation components. The Rhapsody Automotive pack utilizes M2M mapping rules to transform a system logical architecture in SysML into an AUTOSAR architecture, specified in AUTOSAR XML.
MBSE provides a shared, visual understanding of the system, improving communication and collaboration among diverse teams. Modern tools offer features like model versioning, concurrent editing, and integrated communication platforms.
➡️ Authoritative source for system requirements and design
By establishing the model as the authoritative source of system design, MBSE eliminates the inconsistencies and ambiguities often found in document-centric approaches. All design decisions are captured by the model, thereby preventing errors and inconsistencies resulting from siloed documents. As mentioned earlier, model-based digital threads automations ensure unambiguous synchronization of system and component specifications.
➡️ Model management
Effective system model management, including version control, configuration, and access control, is essential to maintain integrity and ensure value throughout the lifecycle, especially across many engineers. Model management shall also facilitate CI/CD pipelines by enabling parallel working, automated merging, and automation APIs. For example, IBM Rhapsody utilizes the Rhapsody Model Manager (RMM) and its API as its model management capability.
➡️ Document generation
In the years to come, documents will be primary work products in enterprise processes, especially among various stakeholders and for process compliance. As of today, documents are key work products made by systems engineers. While the authoritative source becomes the model, automation of standard SE document production, such as SSS, SSDD, IRS, ICD etc., is crucial as part of any modern MBSE environment. It is critical that documents are derived work products and are not maintained as authoritative sources. Most modern MBSE tools provide document generation utilities that streamline the automated production of documents from the model. IBM Rhapsody offers the Publishing Engine (a.k.a “RPE) that supports this important function.
➡️ Model review and markup
Model Review and markup is an important collaborative capability that enables various stakeholders to access and review models by attaching review findings, including actual markups on diagrams. The review and markup are a core collaborative capability that replaces traditional document-based reviews with an online capability on actual model data. SodiusWillert SECollab allows publishing model contents from a myriad of modeling tools, like CatiaMagic, Rhapsody, SparxEA, and more, for online review and markup. SECollab also orchestrates and designates reviewers and tracks the overall review progress.
MBSE shall support iterative development by allowing for rapid model updates and analysis. To achieve this, MBSE tools must integrate effectively with agile planning and tracking tools like Atlassian Jira, similar to traditional coding tools. SodiusWillert OSLC Connect for Jira offers such an integration, including the one featured by IBM Rhapsody.
In addition to agile planning, Model-based testing and automated code generation shall be integrated into Continuous Integration/Continuous Delivery (CI/CD) pipelines, accelerating the delivery of system increments. As mentioned earlier, enabling CI/CD pipelines with MBSE requires a proper set of model management capabilities and automation APIs. That also includes running MBSE tools in “headless” mode to perform various tasks such as:
Here again, IBM Rhapsody is a good example of an integration of an MBSE and code generation tool into a CI/CD pipeline, supporting all the abovementioned necessary capabilities.
The integration of Artificial Intelligence (AI) holds significant promise for further advancing MBSE. Here are some answers to some of the most frequently asked questions.
No. AI can support and enhance MBSE workflows, but it doesn’t replace the need for a structured system model or expert engineering input. Like in software development, AI relies on existing design representations: it augments, not replaces.
AI models work best with text, but traditional MBSE uses diagrammatic formats. SysML v2 introduces a full textual syntax that AI can read and generate, making AI-driven modeling finally practical.
Using SysML v2, AI can produce model elements and fragments based on prompts (much like an "autopilot ») speeding up modeling tasks.
4) Can AI assist with navigating complex models?
Yes. AI can act as a model assistant, helping engineers locate relevant parts and understand the model’s context through natural language queries.
5) What kinds of model analysis can AI perform?
AI can help evaluate completeness, check requirement coverage, suggest optimizations, and identify reusable model components.
Adopting MBSE is not just about deploying new tools. It requires cultural change, investment in training, and a deliberate strategy for integrating model-based practices into established engineering processes. Success requires tackling organizational inertia and demonstrating clear, tangible benefits early on.
Following maturity models, such as INCOSE's MBSE maturity model, helps organizations start with capabilities that are both high-impact and easier to implement. Some teams may prioritize generating interface control documents (ICDs) directly from templates; others may focus on delivering validated use cases to stakeholders.
As organizations move from document-based workflows to model-based development, MBSE improves collaboration, enables early validation, and promotes faster, more traceable delivery of complex systems. With the increasing complexity of systems and the rise of AI-driven tools, MBSE is positioned to remain a solid foundation of digital engineering, supporting traceability, automation, and model continuity throughout the lifecycle.