Engine Failures

Please report Bugs and Problems here
Post Reply
User avatar
DaKurt
FSAirlines Developer
Posts: 452
Joined: Tue Nov 15, 2005 3:49 pm
Location: Munich

Engine Failures

Post by DaKurt » Mon Mar 10, 2008 11:16 am

I added engine failures last night but I'm not sure how often they should occur. Right now every 10 minutes (FlightSimulator-Time) the chance of an engine failure equals the state of the engines. When your engine are at 50% the chance that every 10 minutes an engine fails is 50%.
It seems that an engine would fail to often, but then again I don't think that a real airline would fly with half rotten engines ;)


We have to think about a solution to this problem.
It isn't a very good formula, so if you have a better idea please post it here!
Claudio - FSAirlines Programming

User avatar
CAPFlyer
Chief Pilot
Posts: 3045
Joined: Mon Nov 07, 2005 2:49 am
Location: Lancaster, Texas, USA
Contact:

Post by CAPFlyer » Mon Mar 10, 2008 12:05 pm

Well, several items need to be considered for engine failures -

1) Type of Engine. Pistons have a tendency to fail more often than jets, especially high horsepower radial engines (like the R2800, R3350, and R4630s that power the propliners that are on FlyNET). This is simply because piston engines have more moving parts, thus there are more chances for a failure. Turboprops also fail more often than jets but not as often as pistons.

2) Type of aircraft. Helicopters tend to have more failures than airplanes because of the added vibration and torque that the large rotor places on the aircraft's powerplant systems.

3) Age of the aircraft. The older the airplane (i.e. in or out of production) the more likely that a failure is to occur even on a "perfect" aircraft just due to the natural aging of the parts and the fact that no new engines are being built for the aircraft, only overhauled.

4) Maintenance of the aircraft. The better maintained, the less likely a failure.

5) Fuel quality. The poorer the quality, the more likely a failure because the engine wears out faster trying to process through the bad fuel either in the form of deposits in the injectors and fuel lines & pumps or "hot running" causing the metal to fatigue faster.

6) Engine Handling. The more a pilot abuses an engine, the more likely it is to fail.

While I like the idea of having failures within FlyNET, it's very difficult to quantify the last 2 items reliably and as such hurts the ability to really get a good algorhythm to determine when an engine should fail. The first and second would require new database fields for what kind of powerplant the aircraft has (piston, jet, turboprop) and what kind of aircraft (small, medium, large, helicopter). With the first 4 items in place (#3 is taken care of by the current "in production" status since you can just put a modifier on it that biases against out-of-production aircraft) I think you can come about a fairly robust failure system for all kinds of failures.

Here's what I'd propose given that #1 and #2 are added to the database.

Chance of failure (C)=
Type of Popwerlant (1=jet, 2=turboprop, 3=piston) + Type of Aircraft (1=small, 2=medium, 3=large, 4=helicopter) + Age (1=in production, 2=out of production) + Number of Engines + Maintenance (1=95% to 100%, 2=90% to 94.9%, 3=85% to 89.9%, ect).

Probability of Failure during each hour of flight would be 1% for a score of 5 and then 1% would be added on for each additional point above 5.

The check would be run hourly with a modifier to add 1% to the probability of failure for each hour of flight that has been completed. Thus, on an 18 hour flight, the chance of a perfectly maintained 747-400 to have an engine failure is 26% in the last hour (starting value of 8). The same 18 hour flight on a L-1049 or DC-7 (if they could fly that long) would have a 31% chance of failure (starting value of 13). This seems to me to be fairly realistic and could correspond to not just engines but electrical (generator/alternator) and hydraulic (hydraulic pump or system) failures.
Image

User avatar
DaKurt
FSAirlines Developer
Posts: 452
Joined: Tue Nov 15, 2005 3:49 pm
Location: Munich

Post by DaKurt » Mon Mar 10, 2008 12:27 pm

first of all thank's for writing up your ideas! I like the thought of adding information #1 and #2 to the database in order to create a more realistic failure system. Maybe Konny and Joe could post there opinion about that here, because they are more familiar with database changes.

Of course have to be realistic and cannot use "fuel quality" to calculate the engine status. It has to be a good algorithm which simulates reality quite good, by using as little information as possible. We have to be careful that the system doesn't get to complex...


All in all I like your idea, but a 23% chance of failure on a 18 hours flight would mean that at least every 10th aircraft from munich to singapour (10 hour flight) would have an engine failure. I don't know if this very realistic :?:
Claudio - FSAirlines Programming

User avatar
CAPFlyer
Chief Pilot
Posts: 3045
Joined: Mon Nov 07, 2005 2:49 am
Location: Lancaster, Texas, USA
Contact:

Post by CAPFlyer » Mon Mar 10, 2008 1:59 pm

Not necessarily. If the algorhythm is running on the client and resets for every flight to simulate the "reset" that the engine goes through as it cools down, then the maximum chance that any single flight will encounter is 23%, so in all likelyhood, it'll be more like 1 in every 2300 flights has a failure. The issue here is that when you get out to the edge of the aircraft's endurance, you're really starting to tax the engine's ability to sustain operating temperatures and pressures without failure. The longer the flight, the more likely that the engine is to have problems and need to be shutdown. However, once it cools off, it sort of "resets" itself and then you can go for a while again before the likelyhood of failure increases.
Image

Jos Gielen
Ticket Agent
Posts: 16
Joined: Fri Feb 22, 2008 10:01 am

Post by Jos Gielen » Mon Mar 10, 2008 4:17 pm

With a change of failure of 23% each flight, it is likely to expect an engine failure every 4th or 5th flight!!!!

User avatar
DaKurt
FSAirlines Developer
Posts: 452
Joined: Tue Nov 15, 2005 3:49 pm
Location: Munich

Post by DaKurt » Mon Mar 10, 2008 5:36 pm

yes it is definately too much...


I had another idea of a birdstrike. What about a 0,5% (more or less...we could somewhere look up the real birdstrike numbers) chance of a birdstrike when taxing over 80 knots or you are flying under 100 feet over ground? This might add some realistic and interesting situation while starting and landing
Claudio - FSAirlines Programming

User avatar
joefremont
FSAirlines Developer
Posts: 3694
Joined: Tue May 16, 2006 5:46 am
Location: KSFO

Post by joefremont » Mon Mar 10, 2008 7:30 pm

I think that adding engine type would be useful, aircraft size could probably be implied from the empty weight which is already in the database. In that case maybe we could have an 'Aircraft Type' value that would be Jet/Turbo Prop/Piston/Helicopter.

Personally I think that having the probability of an engine failure at 1 - repair status % ever ten minutes for every engine is much too high. I did some calculations and using that test a four engine aircraft on a eight hour flight with a status of 97%. The chances of getting through the flight without any engine failures is 0.29%, the chance of having all four engines fail is 34.9%.

While I agree that engine/aircraft type should have an impact on how often failures happen I would think it would have more of an impact on how often the aircraft needs to be maintained. I saw a show on TV over the weekend that made the point that one biggest reasons airlines switched from piston engined aircraft to jets was how often the engines needed to be serviced. The average piston engine would need to be reconditioned every 100-200 hours while the average jet could go thousands of hours before needing that level of service.
Image
I've sworn an oath of solitude until the pestilence is purged from the lands.

User avatar
CAPFlyer
Chief Pilot
Posts: 3045
Joined: Mon Nov 07, 2005 2:49 am
Location: Lancaster, Texas, USA
Contact:

Post by CAPFlyer » Mon Mar 10, 2008 7:54 pm

Well, it's not exactly hundreds vice thousands, but pistons had to be serviced more often due to inflight failures more than periodic maintenance unfortunately. The B377 Stratocruiser, DC-7, and L-1049 were all known as "great 3 engined airplanes" because of their unfortunate propensity to have an engine fail inflight, mainly because all 3 flew with high horsepower pistons which had a lot of cylinders and thus moving parts that could break.

The TBO (Time Between Overhaul) on a typical piston engine in airline service was about 1200 Hours, on a typical Turboprop around 2000 Hours, and on a Jet around 3000 Hours. TBO is a "time cycle" set for specific engine types by the manufacturer, but it's very rare to find an engine that ever makes TBO before needing to be pulled and overhauled either because of an in-service failure or out-of-spec part found during inspection.

Anyways, the numbers I used can of course be tweaked if you want simply by moving the decimal point. Instead of whole percentages, have the algorithm divide the result by 10 and then you get single digit percentages and you adjust the modifier for each hour flown to +0.1% and you end up with 2.3% and 3.1% respectively.

Personally, I would say that since we're in a testing mode, we try the algorithm as I proposed originally and if we do get an excessive amount of engine failures, then we adjust it, but I have a sneaking suspicion that failures won't be as often as you think since the algorithm resets for every flight and thus while there is a 19% chance that an engine will fail on any of the 10 hour flights, I think that in practice only about 9% will actually experience a failure if that much.

Joe- We can do that. Per the ICAO (where I got the idea of S,M,L) the definitions are -

Light - 0kg - 17,000kg
Small - 17,001kg - 40,000kg
Lower Medium - 40,001kg - 104,000kg
Upper Medium - 104,001kg - 162,000kg
Heavy - 162,001kg and up

That seems to be a good division of sizes for our purposes.
Image

User avatar
flightsimer
Chief Pilot
Posts: 1815
Joined: Tue Oct 09, 2007 3:35 am

Post by flightsimer » Mon Mar 10, 2008 8:30 pm

only thing i can say what would be wrong is just using if they are in production or not. because there are aircraft that are out there that are still new but arent being produced anymore. example, 717 started in late 90's but no longer is being made, but they are still pretty new.
Owner/CEO
North Eastern Airways

Image
Image

User avatar
CAPFlyer
Chief Pilot
Posts: 3045
Joined: Mon Nov 07, 2005 2:49 am
Location: Lancaster, Texas, USA
Contact:

Post by CAPFlyer » Tue Mar 11, 2008 2:20 am

Okay, so you want to put in a "build date" for every airplane that is currently in the system?

We're trying to keep this simple and aircraft not currently in production are a bit harder to support even if there's a lot of them, a lot of the bigger parts aren't made anymore and all aircraft that are out of production (for some reason) tend to be a bit more prone to breaking, it may not be a whole lot, but it's still there.
Image

vaccaro
Captain
Posts: 488
Joined: Sat Nov 25, 2006 12:03 pm

Post by vaccaro » Tue Mar 11, 2008 3:04 pm

Why trying to complicate things?
If someone wants failure he/she can do it inside FS.
Easy :p

For me this means, start the flight again and again because I never fly with failures enabled, i never fly crash enabled. Besides, if you fly crash enabled, you can crash only once and then leave flynet forever ;).

User avatar
DaKurt
FSAirlines Developer
Posts: 452
Joined: Tue Nov 15, 2005 3:49 pm
Location: Munich

Post by DaKurt » Tue Mar 11, 2008 3:11 pm

the answer is easy. If we do not embed a failure system the mantainance of an aircraft would make no sense. So if you don't repair your aircraft you should know that the probability of a failure is there. If you keep your aircraft in a good condition you should be safe :)
Claudio - FSAirlines Programming

User avatar
Quantum
FSAirlines DB Admin
Posts: 1439
Joined: Sat Apr 29, 2006 8:23 pm
Location: UK

Post by Quantum » Tue Mar 11, 2008 5:10 pm

Hi,

As long as there is a band where no failures will occur then that gives the option to have, or not have failures. Maintain the aircrafts well and no failures will occur. To keep things simple, work purely on the % maintenance state of the aircraft. The lower the %, the chances of a failure are increased proportionately but still randomly. That keeps things simple for the failure systems. If you then want to factor the engine types, then that stays as a seperate calc independant of the failures and affects the rate of attrition of the aircraft maintenance %. This would allow either calculation to be tweaked individually as necessary without a huge rework of a possibly complex calculation.

If VA's wish to accept possible failures then I think a system needs to be developed somehow where a pilot can be rewarded for handling the failures 'professionally'. Not sure how this works as I don't know what is available in FS. One thing that goes against the failure system at the moment is the random gear failure. FlyNET does not allow manual gear extension, stuck means stuck! This would ultimately end in a gear up landing which if crash detection is on results in a 'crash' or if CD is off you will have to back out of the client/flight sim.

Regards

John
CEO - Classic British Flight Services
Classic aircraft on Classic routes
ImageImage

vaccaro
Captain
Posts: 488
Joined: Sat Nov 25, 2006 12:03 pm

Post by vaccaro » Tue Mar 11, 2008 6:13 pm

Good enough!
Then why don't we have an option to do necessary mainteanances instead of a whole A check or similar?
What I try to say is, lets say the hull is perfect, gears are ok but my left engine need mainteanance.
But when I call my technician it takes one day (or longer) for doing the job. Because system thinks that all parts of the a/c has to be maintained.
I am not telling these because don't have many a/c in my hangar. I got plenty and I am the only one who flies them (even though there are 3 other pilots I haven't seen them flying).

Cheers,

User avatar
Jay
Ticket Agent
Posts: 12
Joined: Sat Nov 01, 2008 2:54 pm

Re: Engine Failures

Post by Jay » Sat Nov 29, 2008 11:32 pm

I like the failure thing, its realistic and makes flights more exciting knowing you might have to deal with a failure, although of course its better to keep the engines maintened and not have failures.
I saw someone mention a birdstrike and i think that would be a brilliant feature as then you could have to deal with a problem even if engines are 100%, can other problems happen to, like hydraulic failures and stuff?
also, what happens when you do get an engine failure? if you declare a emergency and manage to land saftley does the pilot get any rewards for doing a safe emergency landing?

happy and safe flying
Image

Post Reply