Thursday, 02 January 2014 16:11




We recently received an email inquiring as to the extent of the failure model on our 737 classic. About the same time, we noticed two separate flight sim forums where members inquired about failure simulation for those products also. In this blog, we discuss IXEG's position on failure simulation, where we are and what we intend to do.

In the real world, almost all failures are unanticipated. Failures are generally brought upon by either faulty workmanship, faulty parts or just some bit of bad luck like flying into a flock of birds. In reality, you might have some excessive wear on a part or a mechanic who got a little sloppy or distracted and in such cases, the crew has no real way of knowing that something might be able to fail. From the perspective of a flight crew, a failure is completely random and unexpected.

In a desktop computer simulation however, the computer only does what it is told, meaning that things are not random at all by default. Randomness has to be purposely programmed into the simulation. Furthermore, for sim enthusiasts, there is really only a narrow range of failures that are relevant or desired. Sim enthusiasts are mostly interested in honing their skills in the more common failure scenarios or challenging their knowledge of the major systems on the aircraft.

The challenge for our team then is to design a failure system that is usable, desirable and configurable for the end user. This means we must pick and choose what we want to fail, how we want to fail it, how the user might want to interact with failure management and also how to configure failures. In short, this means we need to design a dedicated failure management module and GUI interface.

Now x-plane does have a default interface for managing failures; however, because we customize x-plane's model so deeply, only a few of the default failures x-plane provides will work with our 737 where many others will not. The end user has no way of knowing what can or can not be failed on our 737 based on looking at x-plane's failure screens alone. This further necessitates the need for a custom GUI and failures module.

Our strategy is to focus on the systems programming where failure on a component level is integrated into nearly all our virtual components but failures will not be activated when the 737 is initially released. The reason for this is that the underlying core of a failure module has yet to be designed. In our time frame, when given the choice of coding up a working aircraft vs coding a failures module, we obviously must finish the aircraft first before we can choose ways to fail it. Because most users will not utilize a failures module, we find it illogical to withhold release of the product while we develop a failure module; however, by designing component failures into the core code now, we have a quick path to integrating failure management.

After we release the 737, the team can set about designing a failures module that feature a myriad of options like MTBF or explicit failure at user defined points. IXEG is fully committed to systems simulation and failures figures heavily into that commitment. IXEG simply want to make sure it gets done right.

So what might we fail when the time comes? While we cannot say exactly beforehand what we might choose to fail, here is some insight into how we do things and you can extrapolate from there the possibilities. We simulate systems on a component level with components as objects in code. We have switches and knobs of course, electromagnetic relays, hydraulic valves, electrical busses, light bulbs, pumps, generators and inverters, each with their own ability to be failed. We can fail a single annunciator bulb, a single pump, a single electrical bus, a single relay, a single fuel pump, initiate a fuel leak or a hydraulic leak, etc. Failing a generator bus relay alone for example will bring down its entire generator bus and the requisite behavior that ensues. Simply setting a leak rate to a hydraulic tank will cascade into a series of failures as the fluid drops below critical levels, etc. The ability to fail things in our sim exists now. The software to manage those failures does not. We will set about designing a failures module some time after the 737 has stabilized in customers hands and proven itself reliable to users in its normal configuration.


1 comment

Leave a comment

Latest User Comments