[1.8.x] Oh Scrap!- A ScrapYard based Part Failure and Reliability Mod 2.0.1 (07/12/2019)

[Sinu]

I really apologize if this has already been asked and answered but I couldn't find the answer anywhere in this thread:

Do you absolutely need to recover a part to improve its reliability or does simply building it and launching it many times also make it more reliable?

 

[severedsolo]

2 hours ago, Sinu said:

does simply building it and launching it many times also make it more reliable?

This one. They will never be quite as reliable as a part that has had a test flight, but the difference is about 1% on a part that has been built several times.

[severedsolo]

1 hour ago, Oneiros said:

That way, I don't need to know the percentage chance (which I don't understand anyway), or the life expectancy - just the safety rating.

I like this. To be honest, the percentage system has always grated on me a little (my tiny human brain doesn't comprehend why a "10%" failure rate comes up so often, but actually if you run the math, it's pretty likely if you apply that across all parts). The framework for this is already basically in place, would only need minor tweaks. I could possibly also put it into the engineers report (cant be that difficult, other mods hook into it, and I can do away with the horrible "in your face" GUI) Tracking this under #5 - I'll probably leave the percentages as a debug option, because I basically do my balance testing while playing, and having those numbers to hand is good when I'm trying to figure out what I need to tweak.

2 hours ago, Oneiros said:

Also I'm not sure if you're still automatically scrapping parts once they reach their life expectancy, but if so I would suggest leaving it up to the player - so long as they have some indication of the safety level of that part.

With the newer updates the failure rates have been pushed downwards quite heavily, so they are unlikely to be (read: almost never unless you are pushing it to the safety threshold) scrapped. I might make it toggleable, like the "Recover parts above Safety Threshold?" option is already - in fact, I've just realised that option is now redundant, because a part will never go beyond it's safety threshold, so I may re-purpose that option for this instead.

[severedsolo]

On 08/01/2018 at 3:40 PM, Oneiros said:

That way, I don't need to know the percentage chance (which I don't understand anyway), or the life expectancy - just the safety rating.

First pass of safety ratings is now available on the build server - haven't added the report yet, and the GUI will still give you % chances, but it's a start.

[severedsolo]

4 hours ago, Oneiros said:

That was quick! I can't access that site (port could be blocked on my side) but it sounds really good  

I cant access it either. @magico13 - is the server down again?

[magico13]

On 1/10/2018 at 1:29 AM, severedsolo said:

I cant access it either. @magico13 - is the server down again?

Sorry about that, had a power outage happen recently and I either forgot to restart Jenkins or another outage happened but the server restarted. It's up again (and has been since yesterday evening).

[Jade_Falcon]

@severedsolo I noticed that UPFM does not provide Safety Ratings for the PB-X150 and PX50R Pressurized Gass Containers (but oddly the larger versions of these tanks do have ratings).  Is this something that I could change via a config edit? 

Also, I use UPFM along with KerbalHeath, which has Radiation Shielding.  It seems that the shielding is treated as a tank by UPFM, so if a leak develops, all of the shielding is lost.  I dont mind the fact that the shielding can be compromised (I like the challenge!) but in my mind it seems that if shielding failed on a craft it wouldn't completely fail all at once.  KerbalHealth uses pretty small number for the shielding, for most parts its a single digit number, so a failure drains out the shielding pretty fast.  Would it be possible to have UPFM to be able to handle shielding differently than tanks?  My thought would be either a slower failure rate, or just a random percentage loss of the maximum amount (similar to how underthrust works).

Edited February 12, 2018 by Jade_Falcon
Additional information

[severedsolo]

On 11/02/2018 at 10:32 AM, Jade_Falcon said:

 I noticed that UPFM does not provide Safety Ratings for the PB-X150 and PX50R Pressurized Gass Containers (but oddly the larger versions of these tanks do have ratings).  Is this something that I could change via a config edit? 

That's... odd. There is nothing about those two tanks that is special. It should be attaching itself. Having said that, the safety rating feature is unfinished, so I may have just forgotten something. To be clear: is it just not assigning safety ratings, or is UPFM not attaching itself to the part at all? (ie do you see the "trash part" button when you right click it)

On 11/02/2018 at 10:32 AM, Jade_Falcon said:

Would it be possible to have UPFM to be able to handle shielding differently than tanks?  My thought would be either a slower failure rate, or just a random percentage loss of the maximum amount (similar to how underthrust works).

Radiation Shielding is supposed to be on the blacklist. But apparently now I check, I never added it. Whoops. For now you can make a MM patch:

OHSCRAP_RESOURCE_BLACKLIST:NEEDS[UntitledFailures]
{
	name = RadiationShielding
}

 

[Jade_Falcon]

1 hour ago, severedsolo said:

That's... odd. There is nothing about those two tanks that is special. It should be attaching itself. Having said that, the safety rating feature is unfinished, so I may have just forgotten something. To be clear: is it just not assigning safety ratings, or is UPFM not attaching itself to the part at all? (ie do you see the "trash part" button when you right click it)

Radiation Shielding is supposed to be on the blacklist. But apparently now I check, I never added it. Whoops. For now you can make a MM patch:

OHSCRAP_RESOURCE_BLACKLIST:NEEDS[UntitledFailures]
{
	name = RadiationShielding
}

 

Looks like its not attaching, the 'Trash Part' button is missing.

Re: Shielding, thanks, Ill add it!

[fallout2077]

Hey there, folks. This mod looks to be a very awesome addition to KSP, and I intend to use it in my GPP install. However, I was wondering if UPFM conflicts with "Test Flight" by Agathorn, seeing as how both mods are similar in function (albeit UPFM is far more refined and realistic with it's partnership with ScrapYard and Stage Recovery). I'd like to use UPFM in my Realism Overhaul/RP-0 install, however RP-0 requires Test Flight, so throwing that mod away in favor of UPFM is a no-go, unfortunately. Any information would be appreciated, and thank you in advance!

[severedsolo]

On 14/02/2018 at 1:37 AM, fallout2077 said:

Hey there, folks. This mod looks to be a very awesome addition to KSP, and I intend to use it in my GPP install. However, I was wondering if UPFM conflicts with "Test Flight" by Agathorn, seeing as how both mods are similar in function (albeit UPFM is far more refined and realistic with it's partnership with ScrapYard and Stage Recovery). I'd like to use UPFM in my Realism Overhaul/RP-0 install, however RP-0 requires Test Flight, so throwing that mod away in favor of UPFM is a no-go, unfortunately. Any information would be appreciated, and thank you in advance!

Short answer is I don't know, I haven't tried them together.

From UPFM's point of view, it will just ignore what TestFlight is doing and do it's own thing. If TestFlight happens to have failed whatever UPFM is trying to fail, chances are that UPFM will just abort the failure (it does a condition check to check the "working status" is correct before it tries to fail). - This shouldn't affect TestFlight, UPFM will just go "nope can't do that".

 

Edited February 15, 2018 by severedsolo

[severedsolo]

On 11/02/2018 at 10:32 AM, Jade_Falcon said:

 I noticed that UPFM does not provide Safety Ratings for the PB-X150 and PX50R Pressurized Gass Containers (but oddly the larger versions of these tanks do have ratings).  Is this something that I could change via a config edit? 

Just tested this in a near stock game (just UPFM and requirements), and UPFM attached itself properly. My guess would be you have a MM patch thats overriding. You could try adding a :FINAL tag to the TankFailureModule patch - otherwise I will need to see a log.

Edited February 15, 2018 by severedsolo

[severedsolo]

UPFM Beta 15 (0.15) Released

• Further tweaks to the failure model.
• Implemented "safety rating" system - rather than seeing actual failure rates, a part will be assigned a "safety rating" from 0-5.
• The GUI will now display the "safety rating" of your vessel (based on lowest value) rather than percentages. It will also tell you what your worst part is.
• Removed "replace parts" button from GUI as this functionality is now available through ScrapYard
• Fixed a couple of NREs.
• Parts can now exceed the "safety threshold" if they are at End Of Life.
• Space Engines will now be determined by ISP at 1 atmosphere, rather than by thrust (the Poodle was being classified as a launch engine)

Edit: Make that 0.15.1

• Fixed NRE on saving

Edit2: Known issue with 0.15.1 - in flight the GUI will say the Safety Rating is 5, even if it's not. This is largely because that screen was not supposed to be visible from the Flight Scene, but I forgot to turn it off.

On the other hand, maybe I can pass it off as intentional. You cleared it for launch after all, so you must have decided it was safe enough.

Edited February 15, 2018 by severedsolo

[fallout2077]

7 hours ago, severedsolo said:

Short answer is I don't know, I haven't tried them together.

From UPFM's point of view, it will just ignore what TestFlight is doing and do it's own thing. If TestFlight happens to have failed whatever UPFM is trying to fail, chances are that UPFM will just abort the failure (it does a condition check to check the "working status" is correct before it tries to fail). - This shouldn't affect TestFlight, UPFM will just go "nope can't do that".

 

Okay then, that's a relief to know. Does UPFM require a config for a part to work with it, or do parts function out of the box with UPFM?

[severedsolo]

6 minutes ago, fallout2077 said:

Okay then, that's a relief to know. Does UPFM require a config for a part to work with it, or do parts function out of the box with UPFM?

Configs are included, if you find any resources leaking that shouldn't be, there is a MM patch above that will allow you to add them to the blacklist

[fallout2077]

1 hour ago, severedsolo said:

Configs are included, if you find any resources leaking that shouldn't be, there is a MM patch above that will allow you to add them to the blacklist

Alrighty, good to know! I'll have to give it a go when I get home. Thank you very much for the assistance!

[Jade_Falcon]

16 hours ago, severedsolo said:

Just tested this in a near stock game (just UPFM and requirements), and UPFM attached itself properly. My guess would be you have a MM patch thats overriding. You could try adding a :FINAL tag to the TankFailureModule patch - otherwise I will need to see a log.

 

 

I did try the suggestion of the FINAL patch without success (although it didnt seem to beak anything either), so Ill provide some logs to take a look at when you have a moment.

EDIT: Looks like it may be due to Configurable Containers mod, I removed it and those tanks are now tracked by UPFM.  It looks like that mod gave that tank the ability to be switched between Xenon and Argon in the editor using the detail window\Tank Type dialog.  Initially I thought it may have been due to Interstellar Fuel Switch since it also interacts with the taanks, but that doesnt seem to be the case.  The Oscar-B Fuel tanks are also tracked now, but interestingly, they still have their Tank Type edit functionality. 

 

Edited February 16, 2018 by Jade_Falcon
Additional information

[severedsolo]

7 hours ago, Jade_Falcon said:

Looks like it may be due to Configurable Containers mod, I removed it and those tanks are now tracked by UPFM.  It looks like that mod gave that tank the ability to be switched between Xenon and Argon in the editor using the detail window\Tank Type dialog

I wonder if that mod removes the resources from the tank for some reason. :FINAL wouldn't work in that case, because if the resource has already been removed, it will skip over it.

If it's definitely ConfigurableContainers you can probably use either :FIRST or :BEFORE[ConfigurableContainers]

[Jade_Falcon]

13 hours ago, severedsolo said:

I wonder if that mod removes the resources from the tank for some reason. :FINAL wouldn't work in that case, because if the resource has already been removed, it will skip over it.

If it's definitely ConfigurableContainers you can probably use either :FIRST or :BEFORE[ConfigurableContainers]

Looks like :FIRST did the trick, the affected tanks all have safety ratings now (which are quite good as those tanks have been used a *lot*!). Thanks!  And so far, the change seems to play well the ConfigurableContainers, all of the functionality that was there before seems to still work now.

[Micro753]

On 2/15/2018 at 7:06 AM, severedsolo said:

Space Engines will now be determined by ISP at 1 atmosphere, rather than by thrust (the Poodle was being classified as a launch engine)

Instead of changing the failures based on space vs launch engines, why not just make the rate of failures not drop of so quickly and increase the time until failures from age occur? This would keep long missions from being spammed with failures while keeping failures on Mun missions a bit more prevalent. I could see it becoming common practice to leave a craft in orbit for a few months to see if it is going to have delayed failures caused from the stress of launch.

Or get rid of the bathtub curve altogether, or use the bathtub curve only on parts that have been recovered and reused. "Unlike in terrestrial systems, where strain and wear cause failure rate to be a linear function of time, spacecraft failure rates diminish over time. This relationship, a Weibull distribution, is shown in Figure B.1, where the average number of failures reported annually for a given spacecraft decreases."

http://forum.nasaspaceflight.com/index.php?action=dlattach;topic=22274.0;attach=240496

The graph is on page 6. Kind of makes sense too. Most of a spaceship's time is spent doing nothing, not constantly working like an automobile would.

Also, are failures currently possible on part that are turned off? Is they are, please change that. Disabling failures on turned off modules would make redundancy possible. For example, you can put 3 reaction wheels on a probe, but only have 1 turned on. If the one currently on fails, switch to one of the backup ones.

Edited February 17, 2018 by Micro753

[Jade_Falcon]

15 hours ago, severedsolo said:

1 hour ago, Micro753 said:

The graph is on page 6. Kind of makes sense too. Most of a spaceship's time is spent doing nothing, not constantly working like an automobile would. The only real danger to them in space is the solar wind. The parts with the highest chance of failures after being in space should be electronic systems since those will be working more offen and are affected most from the solar wind.

Its true that there is a lot of idle time, but also consider that a device, particularly a mechanical device, that is sitting unused can experience other forms of wear as well, as mentioned on page 10. I wont quote the whole page (its an interesting paper, worth a read):  "...Long periods of storage or transit in space often precede their use.  Mechanical systems can lose lubricant or gather corrosion that leads to later failure."  So the failure model in UPFM may not be too far off;  I've had some long missions that went relatively well (but I do a lot of ground testing and test flights too), and only a few that a critical failure made things unsalvageable, like an underthrust on the only engine that make circularization impossible, or a tank leak on the only tank of the last stage of a launch vehicle. But that makes it a challenge, "space isn't easy."!

One thing that was mentioned in regards to the chart on page 6 and Voyager, they did workarounds and remote fixes to keep the mission going, something that isn't in UPFM.  That might an interesting mechanic that could counter minor failures on long flights.  It wont fix a fuel leak, but might be able to "jiggle" free a stuck actuator on a deployable panel, or compensate for a reaction wheel failure.  But only might, sometimes its just broken.

[severedsolo]

6 hours ago, Micro753 said:

Or get rid of the bathtub curve altogether, or use the bathtub curve only on parts that have been recovered and reused

This is how it already works. A part's likelihood to fail is determined in the Editor (and recalculated on rollout, but 99.9% of the time this returns the same result, although there are a couple of edge cases if you are using KCT where the calc may change) - they don't "wear out" in flight.

6 hours ago, Micro753 said:

Also, are failures currently possible on part that are turned off? Is they are, please change that.

Sort of. A part won't actually fail if it's turned off, but will still roll the dice and if it happens to be turned on at the failure point (or before you switch scene after the failure point) then the failure will happen. I will change this though so it won't even check until the part is switched on.

It's still an interesting read though and I will definitely look through it. As I mentioned earlier up the thread, I tried "realistic" failure rates based on Nasa papers before, and the results were not fun.

Edited February 17, 2018 by severedsolo

[strudo76]

Can you ground test things? For example, put a booster on a launch clamp, and fire the booster without releasing the clamp? Would doing that give an increase to that engine's reliability.

(Might be completely off track. I haven't played with this mod much. I've been one of those who get in an infinite loop waiting for game vs mod versions to stabilise.)

[severedsolo]

Just now, strudo76 said:

Can you ground test things? For example, put a booster on a launch clamp, and fire the booster without releasing the clamp? Would doing that give an increase to that engine's reliability.

You can indeed! ScrapYard will update the "build count" every time a vessel is rolled out (unless you revert, in which case it doesn't count).

If you are intending to recover the parts you are testing for future use, you will need to activate an engine through staging (onLaunch doesn't seem to fire reliably unless an engine comes on - although it does sometimes fire for rover wheels, it's less reliable).

[severedsolo]

So UPFM is I think basically ready to be released. To that end, I will be putting up a Release Candidate very shortly (once I iron out a couple of display issues). It still doesn't have a name though. I personally like "Oh Scrap!" but I have had some good suggestions over the course of the Beta.

So I'll put it to the public. Please take a moment to fill out this short (one question) survey to specify a preference of names. If you have any further suggestions, put them in the "other" field. Also, if I've missed any suggestions (I just searched for "name" in the thread) please let me know.

The survey is here: https://www.surveymonkey.co.uk/r/KWHPW5S