8 Tips on Model-Based Systems Engineering (MBSE)

Published by on April 22, 2014 at 12:55 pm.

tips_mbse_systems

What is Model Based Systems Engineering? According to INCOSE’s SE Vision 2020, “Model-based systems engineering (MBSE) is the formalized application of modeling to support system requirements, design, analysis, verification and validation activities beginning in the conceptual design phase and continuing throughout development and later life cycle phases.”

Systems engineers have been using models for decades to describe technical and operational processes, identify performance requirements, and verify designs. The term MBSE has been around since the 1990s, first used by Vitech as part of their CORE product offering. More recently (2007) a MBSE initiative was launched by the International Council on Systems Engineering (INCOSE) with the goal of making MBSE the norm by 2020.

Many modeling techniques are available, including those found in standards such as the Business Process Modeling Notation (BPMN), ICAM [Integrated Computer-Aided Manufacturing] Definition (IDEF), Systems Modeling Language (SysML), and Lifecycle Modeling Language (LML).

These different techniques are supported by various software tools to make it easier to create the models and track their development, such as MagicDraw, CORE, System Architect, and Innoslate. Ideally, these tools produce results that support the overall lifecycle process from concept through disposal. Although the technique can be drawing using common programs such as Microsoft Visio and PowerPoint, these specialized tools provide more robust databases that enable searching for information needed to make decisions. Many of these tools include simulation capabilities that allow the user to verify the model logic and even derive performance requirements.

So is MBSE right for you and your organization? In general, the answer is a resounding YES! MBSE improves communications by visualizing processes and proposed solutions, improves quality by enabling more consistent documentation and traceability, improves productivity by auto-generation of documentation, and reduces costs by taking rote activities and using the tools to perform them.

However, MBSE adds an extra training requirement: users of the tools and underlying techniques must be trained to use them effectively. In addition, you may have to adjust your current systems engineering processes to best use the tools. You also have to be careful not to just “feed the tool.” Garbage in- garbage out still applies. And you will want to be careful not to drive the models to low a level, thus wasting time on irrelevant detail.

To avoid problems with applying MBSE, here are 8 tips that will help you.

  1. Define the system context. The system context: “describes the system relationships and environment, resolved around a selected system-of-interest” [SE Body of Knowledge]. This context defines the “top” from which you will decompose the requirements, actions and assets that make up the system. This context is essentially the scope of your design or project.

  2. Document the assumptions. Since any model is an approximation of reality (it takes the universe to completely model the universe), the assumptions of what you include in the model and as importantly, what you exclude are critical.

  3. Make sure to identify a broad set of scenarios or use cases. As part of the analysis, identifying the potential uses of the systems, including maintenance, it very important. Often a set of scenarios are developed that overlap or miss ways the systems may be operated and maintained. These scenarios are often developed as part of the Concept of Operations (CONOPS).  Make sure you model all the scenarios and include people as resources.

  4. Verify your model as you develop it. Some of the modeling tools include a built-in simulation capability. These capabilities vary from simple model walkthroughs to Monte Carlo simulations. The strong the simulation capability, the better the model logic will be. For example, Monte Carlo provides the opportunity to automatically explore all levels and decisions in the model, discovering problems long before specifications are developed. By doing this you reduce risk, help ensure quality, and reduce long-term costs.

  5. Know when to stop decomposition. You can start at a high level and decompose a model down to a very low level of activity. But is this always useful? When should you stop? I now use a couple of rules of thumb for this. I will decompose three levels down across the model and then stop to see when I am. If I am close to being able to specify a component that I can buy off the shelf or easily build, then I stop completely. Another technique is what we call “middle-out.” We model the various operational (and maintenance) scenarios) and use that to abstract up to create the high level requirements and then decompose down to identify systems and components.

  6. Calibrate your model. Models of real systems and processes include people. Hence, they are “non-linear” mathematical systems. As a result, if you try to extrapolate beyond where you have data to calibrate the model, the model will lie to you. Make sure when you model anything, you have data to support the results. Work with subject matter experts to make sure the models predict things they expect, and if not, why not.

  7. Support the V&V process. The models need to support the entire lifecycle, but especially Verification and Validation. (V&V). The models provide a basis for the more detailed simulations that are part of V&V. Make sure the models and requirements trace to the analyses, simulations, and tests that you plan to conduct. You can use the results of the V&V process to further refine and calibrate the model for the next time you need it.

  8. Use a MBSE software not just word processing, drawing and spreadsheet tools. Microsoft Office and similar products are wonderful for developing documents. That’s what they were designed to do. But to create robust models you need a modeling tool. A number are out there, including MagicDraw, CORE and System Architect. We recommend Innoslate as a new cloud-based tool that reduces costs and provides more complete capabilities for modeling and simulation.


Topics: ,