inner banner

Diagnostic Optimization of Large-Scale Systems

by DSI Staff | Published 06/28/2004 | Concepts

Scalability of the System Engineering Process

One of the foremost challenges in large system design is keeping the engineering effort scalable. This is typically where engineering software becomes invaluable. As such, the question becomes, “How scalable is the software you are using?”

DSI has always recognized this problem as one of the principle design goals of the eXpress software. It is fairly easy to identify some key areas in which the system size becomes a concern:

  • Introducing additional lower-level data
  • Changing connectivity (model structure)
  • Changing testing and diagnostic approaches
  • Validating designs against the specification or other models

Solving the scalability problems associated with each of these provides the ability to run assessment reports so that optimization can occur.

Introducing additional lower-level data

The introduction of lower-level data occurs throughout the engineering process. Sometimes the introduction is due to new data becoming available from a subcontractor, while other times it is because the top-down design process is pushing for more information. Regardless of the reason, the ability to quickly add lower-level data to an existing, self-consistent set of models is critical.

The solution is a familiar term, frequently over-used, but not as common in tools as one might hope–Object-Oriented Design. A tool that provides a good object-oriented approach allows the addition of lower-level detail to override a high-level definition.

In eXpress, this is accomplished by linking new designs to existing objects. What eXpress does that is unique is that it retains three modes of hierarchical operation. First, it can switch the relationship to a bottom-up mode, if it is desirable for the lower-level model to be considered as the sole source of information. Second, it can alternatively use a top-down mode so that the top-level can require a particular approach from the lower-level. In this way, it serves as an ongoing requirements flow-down, until such time as the lower-level complies and the system stops changing. Third, a hybrid mode is supported so that some information can flow down, while other information can flow up.

The scalability offered in such an approach is three-fold:

  • Early analysis for estimation purposes does not require lower-level data.
  • Mixed modes of hierarchical relationships allows convergence towards a solution.
  • Mixed modes of hierarchical relationships support different ways of working with subcontractors.

Assessing the engineering tools scalability to these types of problems is an important step in ensuring the ability optimize throughout the engineering process.

Changing connectivity (model structure)

Another factor that can degrade a tools ability to support optimization throughout development is the degree to which connectivity changes can be made without disturbing other definitions. Frequently, as the connectivity changes, other layers of information are essentially destroyed, causing additional re-engineering work that slows the turn-around time, or even prevents a response in the timeframe it is required altogether.

If turn-around time in a tool reaches the point where the turn-time exceeds the window of opportunity in which recommendations can be accepted, the tool fails to take part in the optimization process.

This design consideration was once again factored into how eXpress manages its information. Unlike any other approach, eXpress supports more than a simple layering of data. Instead, it uses a layering and abstraction approach so that, for example Test Definitions tolerate connectivity changes, so that they can automatically update in response to it. This is a leap beyond other approaches in that connectivity changes otherwise require Testing to be re-entered and re-analyzed. eXpress incorporates topological abstraction into its approach, thereby eliminating virtually all Test rework associated with connectivity changes.

Of course, approaches like this must extend throughout the tool, as they do in eXpress. Like the old addage, software tools as well, are only as good as their weakest link.

Changing testing and diagnostic approaches

Another aspect of the design that can change rapidly and often is the testing and diagnostic approach itself. As changes are made to the system, due to improvement for maintenance purposes, reduction of critical behavior, elimination or control of feedback loops, etc., the testing approach typically changes in response.

Since assessment hinges on an accurate representation of how the system will be diagnosed, it is critical to be able to rapidly configure various testing approaches. In fact, the Testability or Systems Engineer will often try out different testing scenarios to kick start the optimization process.

eXpress approaches this problem by allows flexible topological abstraction as a core of its testing approach. Engineers can define test coverage using high-level definitions that relate to the topology. This not only abstracts the topology, making it stable against topological changes, but also reduces the time usually spent manually determining fault propagation for the purpose of calculating test coverage. Going further, eXpress supports hybrid definitions–that is, both functions and failure modes can form parts of the definition. This means that early functional testing can give way to later failure mode testing, without destroying traceability or convergence towards the ultimate solution.

Rapid Diagnostic Assessment, whose foundation is accurate test definitions, is another essential factor in determining scalability.

Validating designs against the specification or other models

Finally, we come to validation. Validation is too often overlooked as a job that falls on the design side of the fence. However, validation is essential to any model to ensure that it matches the design itself.

When it comes to large-scale systems, the design changes also become larger scale. As such, one can not afford to spend time both on updating the model to match the changes as well as revalidating the entire system. This drives the need for the engineering software to support many methods of validation that are easily compared against previous validations and engineering baselines.

eXpress’ solution to the validation problem is multi-faceted. First, keeping the engineer on track throughout their change process is done through an integrated error-checker. This helps notify the engineer immediately of definition problems before time is wasted solving problems that show up on dependent layers of information. Second, eXpress provides the most extensive reporting available. Reports have subreports, which have subreports, etc. This allows data drill-down into any aspect of the design.

It is the most subtle of restrictions in a tool that make it less scalability. Unfortunately for many, these restrictions are only found after significant investment into a particular tool or methodology. As a result, thorough research and a proven track-record is critical in tool selection. The superficial features that are inherent to any group of tools that attempt to solve a problem are not the ones that will determine its ultimate success.

Subscribe To Our Newsletter