Data Engineering for Model-Based Systems

Engineering:
Powering the Future of Complex System Design
Fundamentals, Applications, and Modern Technologies

Prof. DrEng. Ir. Taufik Djatna, MSi

Full Professor in Industrial System Engineering

Department of Agro-Industrial Technology
IPB University
 OUTLINE & OBJECTIVES
Introduction:
• The Convergence of Systems and Data
Explain the role
Fundamentals: and
• What is MBSE? What is Data Engineering? Define MBSE importance of
and Data Data
The Intersection: Engineering. Engineering in
• Why DE for MBSE? Scope & Value the MBSE
lifecycle.
Core Concepts:
• Digital Thread, MBSE Data Types, Data Pipelines

Enabling Technologies:
Identify key Describe
• Storage, Processing, Analytics types of data fundamental DE
Challenges & Solutions: generated and concepts
consumed in applied to MBSE
• Current Hurdles and Innovations MBSE. data.
Case Study Snippets:
• DE & MBSE in Action

Future Outlook & Key Takeaways

THE CHALLENGE: INCREASINGLY COMPLEX SYSTEMS
: WHY DO WE NEED BETTER APPROACHES?
• Modern systems (aerospace, automotive, energy,
healthcare) are incredibly complex.
• Interconnected components (hardware, software,
Problem networks, humans)
• Stringent requirements (safety, performance,
Statement: security, sustainability)
• Multi-disciplinary teams, geographically distributed
• Rapid technological evolution & shorter
development cycles

Traditional • Struggles with complexity, consistency,

Document- communication, and traceability. Leads to errors,
delays, cost overruns.
Based SE:

• "The complexity of systems has grown faster than

Quote: our ability to manage it using traditional methods."
(Paraphrased common sentiment)
 WHAT IS SYSTEMS ENGINEERING (SE)? : THE FOUNDATION: SYSTEMS
ENGINEERING

• "Systems Engineering is an
interdisciplinary approach and means
to enable the realization of successful
systems. It focuses on defining
Definition customer needs and required
functionality early in the development
(INCOSE): cycle, documenting requirements, then
proceeding with design synthesis and
system validation while considering
the complete problem..."

• Requirements Engineering, Architecture Design,

System Analysis, Integration, Verification &
Core Activities: Validation (V&V), Project Management.

• To manage complexity and ensure the system

meets its objectives throughout its lifecycle
Goal: (concept to disposal).
 INTRODUCING MODEL-BASED SYSTEMS ENGINEERING (MBSE): THE
EVOLUTION OF MODEL-BASED SYSTEMS ENGINEERING (MBSE)
•"MBSE is the formalized application of modeling to support
Definition system requirements, design, analysis, verification, and
(INCOSE validation activities beginning in the conceptual design
phase and continuing throughout development and later
MBSE Wiki): life cycle phases."

•From document-centric to model-centric. The model

Shift: becomes the primary artifact and source of truth.

• Improved communication & stakeholder understanding

• Enhanced ability to manage complexity
• Improved system quality & reduced errors (early detection)
Key Benefits: • Enhanced knowledge capture & reuse
• Stronger traceability between requirements, design, and
V&V

Common
Languages/ •SysML (Systems Modeling Language), UML (Unified
Notations: Modeling Language), UPDM, Arcadia/Capella.
 INTRODUCING DATA ENGINEERING (DE): THE ENABLER
• Data Engineering is the discipline focused on
the practical application of data collection,
storage, processing, and analysis. It involves
Definition: designing, building, and maintaining the
infrastructure and systems that allow
organizations to handle and utilize large
volumes of data efficiently and reliably.

• Data Ingestion (Collecting data from sources)

• Data Storage (Databases, Data Lakes,
Warehouses)
• Data Transformation (ETL/ELT processes -
Clean, Aggregate, Enrich)
Core Activities: • Data Pipeline Orchestration (Managing data
flow)
• Data Governance & Security (Quality, Access
Control, Compliance)
• Building systems for Data Scientists &
Analysts

• Make quality data available, accessible, and

Goal: usable for downstream applications (Analytics,
ML, Reporting, Operations).
 THE DATA EXPLOSION IN ENGINEERING & MBSE : WHERE
DOES THE DATA COME FROM IN MBSE? - MBSE GENERATES DIVERSE DATA:
• Blocks, Interfaces, Ports, Requirements, Use Cases, Activities, State Machines,
Model Elements: Parameters. (Structured/Semi-structured)

Relationships: • Dependencies, Allocations, Traces, Connections. (Graph-like structure)

Simulation & Analysis • Performance metrics, Failure modes, Safety analyses, Trade-off studies. (Numerical,
Results: Tabular, Time-series)

V&V Data: • Test cases, Test results, Coverage analysis, Issue tracking. (Structured, Textual)

Configuration • Model versions, Baselines, Change logs. (Metadata)

Management Data:

Process Metadata: • Who changed what, when, why? Collaboration logs.

•Requirements from external sources, operational data for calibration, component

MBSE Consumes Data: libraries, legacy system data.

•Volume, Velocity (simulations, real-time monitoring), Variety (different formats,

Challenge: tools), Veracity (consistency, accuracy). -> A Big Data Problem!
THE INTERSECTION: WHY DATA ENGINEERING FOR MBSE?
BRIDGING THE GAP: DE MEETS MBSE
The Synergy: Key Value Propositions: Core Idea:
1.Scalability: Handle large, complex
• DE provides the system models and associated • Treat MBSE
foundation to data (simulations, tests).
artifacts and
2.Integration: Connect MBSE data
manage and with other enterprise systems (PLM,
related lifecycle
leverage the data ERP, ALM, MES, Operational Data).

generated and
Creates a holistic view.
3.Analytics: Enable advanced
data as first-class
consumed by MBSE analytics on system models and data assets
lifecycle data (e.g., impact
processes at scale. analysis, pattern detection, requiring robust
predictive maintenance insights).
4.Data Quality & Governance: engineering
Ensure consistency, reliability, and
traceability of MBSE data across
practices.
tools and teams.
5.Collaboration: Provide a robust
data backbone supporting
collaborative modeling and
analysis.
6.Automation: Automate data flows
for reporting, V&V, and model
updates.
 SCOPE OF DE ACTIVITIES IN THE MBSE CONTEXT
WHAT DOES A DATA ENGINEER DO HERE? → SPECIFIC TASK:

Designing Data Architectures:

• Choosing appropriate storage (Graph DBs? Relational? Data Lakes?) for model data, simulation results, etc.
Building Data Pipelines:
• Ingesting data from various modeling tools (e.g., Cameo/MagicDraw, Rhapsody), simulation tools
(MATLAB/Simulink), test systems.
Data Transformation:
• Standardizing data formats (e.g., using OSLC, STEP, or custom schemas), linking related data elements
across domains.
Implementing Data Lakes/Warehouses:
• Creating centralized repositories for integrated engineering data.
Ensuring Data Quality:
• Implementing checks for model consistency, completeness, and adherence to standards.
Setting up Analytics Platforms:
• Providing tools (e.g., Spark, Python libraries) for querying and analyzing MBSE data.

Managing Data Access & Security:

•
• Controlling who can view or modify sensitive design data.
CORE CONCEPT 1: THE DIGITAL THREAD
--- WEAVING IT ALL TOGETHER: THE DIGITAL THREAD

MBSE & The Role of Data

Definition: Digital Thread: Engineering:

A data-driven architecture linking MBSE models form a critical part DE provides the infrastructure and
information generated throughout of the Digital Thread, representing integration mechanisms to build
the product lifecycle, connecting the authoritative source of truth for and maintain the Digital Thread.
processes and enabling a holistic system design and requirements.
Connecting disparate data silos
view of the asset's data (from (CAD, CAE, PLM, MBSE, ERP, MRO).
concept to operation and
disposal). Ensuring data traceability and
consistency across the thread.
Making Digital Thread data
accessible for analysis and
decision-making.
 CORE CONCEPT 2: UNDERSTANDING MBSE DATA
-- WHAT KIND OF DATA ARE WE HANDLING?

• Highly Connected: Relationships between model elements are

crucial (graph-like).
• Heterogeneous: Different types (structural, behavioral,
requirements, parametric) and formats (XML/XMI from tools,
Data simulation outputs, test logs).
• Versioned & Configurated: Systems evolve; tracking changes
Characteristics: and variants is critical.
• Context-Dependent: The meaning of data often depends on its
place within the model structure and lifecycle phase.
• Often Requires Domain Knowledge: Understanding SysML/UML
and the specific system domain is vital.

• Storage choice needs to handle relationships well (Graph DBs are often
suitable).

DE Implications: • ETL/ELT needs to parse complex formats (e.g., XMI) and maintain
semantic integrity.
• Data models need to accommodate heterogeneity and versioning.
 CORE CONCEPT 3: DATA PIPELINES FOR MBSE MOVING
AND TRANSFORMING MBSE DATA
• Automate the flow of data from MBSE tools and
Purpose: related sources into centralized storage and
processing systems.

1.Extraction: Pulling data via APIs from MBSE tools (e.g.,

Typical Stages: Cameo Collaborator REST API), file exports (XMI),
simulation output files, test databases.

1.Parsing complex structures (XMI, proprietary formats).

2.Standardizing schemas/ontologies (e.g., mapping tool-
specific elements to a common model).
Transformation: 3.Linking data across domains (e.g., linking a
requirement ID to a model element ID and a test case
ID).
4.Data quality checks (validation rules).

1.Storing processed data in the target system (Data

Loading:
Lake, Graph DB, Warehouse).

• Apache NiFi, Kafka, Airflow, Cloud-specific services

Technologies: (AWS Glue, Azure Data Factory, GCP Dataflow),
Python scripting.
DETAILED VIEW OF AN MBSE DATA PIPELINE SHOWING SPECIFIC TOOL INPUTS AND
TRANSFORMATION STEPS
STORING MBSE DATA: LAKES, WAREHOUSES, & GRAPH DBS
WHERE DOES THE ENGINEERED DATA LIVE?
• Concept: Store raw, diverse data in its native format. Schema-on-read.
Data Lakes: • Use for MBSE: Staging area for raw tool exports, simulation logs, unstructured text.
Good for exploration and future-proofing.
• Technologies: AWS S3, Azure Data Lake Storage (ADLS), GCP Cloud Storage, HDFS.

• Concept: Store structured, processed data optimized for reporting and BI. Schema-
on-write.
Data Warehouses: • Use for MBSE: Storing aggregated metrics, KPIs, historical trends derived from models
and tests.
• Technologies: Snowflake, BigQuery, Redshift, Synapse Analytics.

• Concept: Optimized for storing and querying highly connected data (nodes, edges,
properties).
Graph Databases: • Use for MBSE: Excellent fit for representing model structure, relationships, and traceability.
Enables powerful pathfinding and impact analysis queries.
• Technologies: Neo4j, Amazon Neptune, Azure Cosmos DB (Graph API), TigerGraph.

Hybrid Approaches: • Often a combination is best (e.g., Lake for raw data, Graph DB for model structure,
Warehouse for reporting)
 DATA GOVERNANCE & QUALITY IN MBSE
-- Ensuring Trustworthy Engineering Data
Why Critical? Key Aspects: Challenge:
• Decisions based on • Data Quality: Accuracy, Completeness, • Applying governance
Consistency (across tools/models), Timeliness,
models (design Validity (conformance to standards/rules). across heterogeneous
choices, safety V&V) • DE Role: Implement automated quality checks tools and processes.
have high stakes. Errors in pipelines, data profiling.
in data can lead to • Metadata Management: Documenting data
system failures. sources, definitions, lineage (how data was
transformed).
• DE Role: Implement metadata catalogs (e.g.,
Apache Atlas, Alation, Collibra).
• Data Lineage: Tracking data from origin
through transformations to consumption.
Crucial for impact analysis and debugging.
• DE Role: Tools and pipeline design should
capture lineage.
• Master Data Management: Defining and
managing authoritative sources for key entities
(e.g., standard components, requirements).
• Access Control & Security: Protecting sensitive
intellectual property and ensuring compliance.
Analytics On MBSE Data: Unlocking Insights --
Beyond Modeling: Analyzing The System Data
Example Analytics Use
Goal: Technologies:
Cases:
Impact Analysis: "If I change this requirement or component, what other parts of the
system (design, tests, documentation) are affected?" (Graph queries are powerful here).

Leverage the Model Completeness & Consistency Checks: Automated checks beyond basic tool SQL, Cypher (for
validation (e.g., identifying orphaned elements, inconsistent naming).
integrated MBSE Graph DBs),
data for deeper Requirements Coverage Analysis: Automatically verifying which requirements are covered
by design elements and test cases. Spark SQL/MLlib,
understanding Python (Pandas,
System Optimization: Analyzing simulation results across many runs to find optimal
and better design parameters. Scikit-learn), BI
decision- Defect Prediction: Using historical data (model changes, test results, issues) to predict tools (Tableau,
making. areas prone to defects.
Power BI).
Design Pattern Recognition: Identifying recurring architectural patterns across projects.
Current Challenges & Problems : The Hurdles We Face
• Different MBSE/Engineering tools use proprietary formats; getting data out and integrating it is
Tool Interoperability: hard. Standards like OSLC (Open Services for Lifecycle Collaboration) help but aren't universally
adopted or sufficient.

• Tools may represent similar concepts differently. Mapping them requires domain expertise and
Semantic Heterogeneity: robust semantic mediation.

• Analyzing very large system models (millions of elements/relationships) can challenge even
Scalability of Graph Queries: specialized graph databases.

• Ensuring consistency and accuracy across a fragmented toolchain is difficult. Lack of standard
Data Quality Assurance: validation rules.

Versioning & Configuration • Handling evolving models and their associated data consistently across the DE platform.
Management:

• Systems Engineers and Data Engineers often have different backgrounds and terminology. Effective
Bridging Cultures: collaboration is key.

• Protecting valuable IP contained within models when data is centralized and integrated.
Security:
 Cloud
Low-
Platforms Improved
Graph Knowledge AI/ML Data Mesh Code/No-
(AWS, APIs &
Databases: Graphs: Integration: Concepts: Code
Azure, Standards:
Platforms:
GCP):

Provide
scalable
Combining
storage (Data
graph
Lakes), Applying ML for Decentralized
Maturing databases with
processing predictive approach to Emerging
rapidly, offering ontologies Tool vendors
(Spark, analytics, data platforms aim
better (semantic are slowly
Serverless anomaly ownership and to simplify
performance models) to improving APIs;
Functions), detection in pipelines, pipeline
and features handle ongoing work
databases simulation/test potentially
(Managed
specifically for heterogeneity
data, Natural aligning well
creation and in standards
connected and enable data bodies (OMG,
SQL, NoSQL, Language with distributed
data relevant more intelligent integration for INCOSE, ISO
Graph), AI/ML Processing engineering
to MBSE (e.g., querying. domain STEP
services, and (NLP) on teams owning
Neo4j, (Related to experts. AP233/242).
orchestration requirements their domain
Neptune). Semantic Web
tools. Form the documents. data.
tech: RDF,
backbone of
OWL, SPARQL).
modern DE
solutions.

Modern Solutions & Technologies

Tools And Approaches Making Headway
 Example 1: Aerospace - Example 2: Automotive -
Impact Analysis ADAS V&V Data Integration
Scenario: Integrating simulation data (e.g., from
Scenario: Assessing the impact of changing a thruster CarMaker/Simulink) and physical test drive data with MBSE
specification on a satellite system model. requirements models for Advanced Driver-Assistance Systems
(ADAS).

DE Role: Ingested SysML models (e.g., from Cameo) into a Neo4j DE Role: Built data pipelines using Kafka and Spark to ingest,
graph database. Developed Cypher queries to traverse the clean, and correlate time-series sensor data, simulation logs,
model graph (requirements -> functions -> logical blocks -> and test results. Stored in a Data Lake. Linked results back to
physical components -> V&V cases). requirements traced in the MBSE model (e.g., Rhapsody).

Outcome: Rapid identification of all affected elements, Outcome: Enabled comprehensive coverage analysis,
associated tests, and documentation, reducing analysis time automated report generation for safety standards (ISO 26262),
from days to minutes. and faster debugging of failed test scenarios.

Case Study Snippets: DE & MBSE In Action

: Real-world Examples (Use 1-2 Brief Examples)
 Native cloud-based MBSE platforms with built-
Tighter Tool Integration: in DE capabilities. More seamless API-driven
Future Trends
data exchange.
& Research
Using AI/ML directly within the modeling process (e.g., Directions
AI-Powered MBSE:
suggesting design alternatives, auto-generating parts
of models from requirements, AI-driven simulation). DE What's Next?
provides the data foundation for this.

MBSE models forming the core of operational Digital

Twins, continuously updated with real-world data via
Digital Twins: DE pipelines. Requires robust streaming data
capabilities.

Making MBSE-derived data accessible to more

Democratization of Data: stakeholders beyond core SE teams via user-
friendly analytics interfaces (powered by DE).

Integrating modeling environments with data

Enhanced Collaboration Platforms: platforms for real-time, data-aware collaboration.

Continued push for interoperability standards

Standardization Efforts: (OSLC, SysML v2 aims for better API support).
 CONCLUSION & RESOURCES : KEY TAKEAWAYS &
FURTHER LEARNING

Call to Key
Summary:
Action: Resources:
INCOSE (International Council on Systems
MBSE is transforming systems engineering by using
Engineering): www.incose.org (MBSE initiatives, SE
www.incose.org

models as the central artifact.

Body of Knowledge)

MBSE generates and consumes vast amounts of Object Management Group (OMG):
complex, connected data. www.omg.org (SysML Specification)
www.omg.org

As future engineers, understanding both the Relevant Journals/Conferences: INCOSE

Data Engineering provides the essential methods
system modeling domain (MBSE) and the data Symposium, IEEE Systems Journal, ACM
and technologies to manage, integrate, and
handling domain (DE) will be increasingly SIGMOD/VLDB (for DE aspects), Complex Systems
analyze this data at scale.
valuable. Design & Management (CSD&M).

The synergy between DE and MBSE enables the

Cloud Provider Documentation (AWS, Azure,
Digital Thread, advanced analytics, and more
GCP) on Data Engineering services.
robust system development.

Books on Data Engineering (e.g., "Fundamentals

Key technologies include Cloud platforms, Graph of Data Engineering" by Reis & Housley) and MBSE
Databases, Big Data processing, and AI/ML. (e.g., "A Practical Guide to SysML" by Friedenthal,
Moore, Steiner).
 Q&A: •</\>

Contact •[email protected]
Info •H205 –FATETA Building