Model Based Risk Analysis
In many cases, risk analysis is equated with the process of filling out an FMEA form sheet. But soon people find out that this approach is part of the problem and by no means a solution.
Even if the system or process under consideration is only moderately complex, and the form is not longer than a few pages, it becomes more difficult and in the long run it eventually becomes virtually impossible to produce a complete and consistent risk analysis.
Studies which we performed more than 20 years ago, when we started research for a unique way of performing risk analysis showed us, that people tended to confuse causes and effects, could not communicate their results to other people who were not members of the team that initially produced the form and did not understand their own form after a few weeks had passed.
Sound reasoning about failures in a system is only possible when the workings of the system are sufficiently understood by all members of the team. This means that risk analysis must be based on an analysis (or a model) of the system/process under consideration which is as abstract as possible and as specific as necessary. Whenever a failure is introduced, the team must be absolutely clear that it knows which function of the system fails. Also, the physical component which produces the function must be known.
The model which is the basis of APIS IQ-Software ensures that:
- for every failure the connected function is known
- for every function, the system or process element which produces this function is known
- the structure of the system or process under consideration is known
APIS IQ-Software has specialized tools (called "editors") for every step of the analysis process. These tools are powerful as well as easy to use and they guarantee the consistency of the mode as required by the US AIAG (which is the equivalent to the German VDA). As this approach is in no way specific to the automotive industry, APIS IQ-Software also has a constantly growing user base in other industries, notably the Aerospace, Defence and Alternative Energy and Health Industries.
All tools of APIS IQ-Software operate on one unified model, without redundancy, that guarantees that an item of information is always up to date in all different views (forms) which the software supports, no matter where it is changed. So, if for example you change the process step number of a process step in the process flow diagram, this information will be immediately reflected in the control plan and in the FMEA form without any additional work.
VDA recommends that the analysis is performed in several steps; the following table gives an overview of the required phases and the tools of APIS IQ-Software that support this step.
| Phase | Goal | supported by |
|---|---|---|
| Structure Analysis | Identify the physical components of the product or process ("System Elements") that produce the desired functions as well as their interdependence | Structure Editor Function- and Failure-Analysis (FMEA form) |
| Functional Analysis | Identify the functions of the system elements and their interdependence | Structure Editor Function- and Failure-Analysis Function Net Editor (FMEA form) |
| Failure Analysis | Identify the failures of the functions and their interdependence | Structure Editor Function- and Failure-Analysis (Function Net Editor) Failure Net Editor (FMEA form) |
| Risk Evaluation and Optimization | Identify the critical failures and define Action(s) to reduce overall system risk | FMEA form Structure Editor Function- and Failure-Analysis Function- and Failure-Analysis Failure Net Editor |
| Reporting and Controlling | Statistical analysis and presentation of the risk; controlling if the implementation of actions | Deadline Editor Statistics Editor |
| Data Mining | Reuse of knowledge ("Lessons learned") | IQ-Explorer |
APIS IQ-Software does not require that these steps are performed in the order given above. However, this is highly desirable. Instead, it supports every process model (top down, bottom up or middle out) while always ensuring the consistency of the data. Just to name an example: you can have the software construct the function net based on a given failure net!