Software Review: How complex is that component? Software tells all
Edited by Leslie Gordon
We use the software to design and analyze missioncritical components such as adapters that connect rocket launchers to their payloads. Comprehensively measuring complexity provides a quantity that can then be used as a design attribute (just like mass, frequency, or stress) to develop products that are easier to build, assemble, service, and repair.
Engineers start by setting up and running a Monte Carlo simulation (for probability) in programs such as iSight, LMS-Optimus, Ansys-PDS, PAM-Opt, or VeroSolve. (OntoSpace can also process data from tests, sensors, and historical records.) Statistics of input variables, such as loads, dimensions, and material properties, determine factors such as a component’s probability of failure. Results are in the form of rectangular tables of data, which are imported into OntoSpace. Basically, the software postprocessors the results and extracts knowledge from them.
Some of this is shown in what is called the Process or Knowledge Map. It provides a new way to see information and understand how dynamic systems function. It illustrates which design variables affect performance. It also highlights weak points as well as “hubs,” i.e., variables crucial to system functioning, and lets users verify if designs are redundant.
The software calculates the total amount of entropy (a measure of disorder or randomness in a system) and uncertainty (the estimated amount by which a calculated value might differ from its true value) and uses this information to quantify the system’s complexity and also provide a measure of its robustness — the capability to maintain functionality in the presence of internal or external disturbances.
Another tool, The Navigation Table, lets engineers establish constraints and objectives, and verify if a system can perform required functions and at what level of risk. The tool is helpful in understanding combinations of inputs that will cause certain effects in outputs. An important feature detects what are called outliers. These represent pathological circumstances — unlikely events — that often lead to failure. The feature is key to assessing risk in mechanical components. No other CAE software performs this kind of analysis. Because OntoSpace is not based on statistics, it can handle data of any kind, even if ill-conditioned, that is, in principle, suitable for the intended purpose but in some way distorted or incomplete. In the past, using complexity as a design goal or attribute was not possible. OntoSpace technology makes it possible to imagine complexity- based CAD. The concept:
Given two or more equivalent designs in terms of cost, function, and performance, it’s always better to select the one with the lowest complexity. This is something the developer should look into. As components become increasingly complex, it is important to include complexity in the design loop from the start.
A downside is the software currently reads only comma separated values or plain text files. It would be great if the software could directly read Excel data.
The software is easy to use. Execution times for typical problems are on the order of minutes, sometimes seconds. However, running Monte Carlo simulations might be a problem for many engineers. The developer provides a tutorial on this, and documentation is generally complete. It also provides a two-day training course on complexity-management technology. E-mail support is efficient.
Users can upload a rectangular table of up to 100 variables and 1,000 samples to the developer’s online service and have it analyzed in a matter of minutes. Cases of up to 10 variables are analyzed for free. OntoSpace comes from Ontonix Srl, Via Lega Insurrezionale 7, 22100 Como, Italy, +39-031-3100059, ontonix.com
—Jorge Vilanova
A Process Map for a structural component shows inputs and outputs indicated with different colors. Intense red and blue nodes represent, respectively, input and output hubs.
The situation indicated in the Navigation Table has a probability of occurrence of 4%. The user defines constraints and objectives in fuzzy terms (that is, high, very high, medium, low) and the tool establishes which combinations satisfy user requirements and with what probability.
A scatter plot illustrates several outliers (shown in orange). Similar data cannot be treated with traditional statistical techniques.