Jump to content
  • 82

Broken DeltaV with multiple engines and/or with side tank attached [fuel tank placement order messes up dV calculations]


Niko_Dragovich

Bug Report

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 11 Pro | CPU: i5 13600KF | GPU: RX 7900 XTX | RAM32GB DDR5 6000 CL32

 

Hi.

It's very complicate for me to explain this bug. I'll try to explain it with screenshots in the video below. When I have 2 engines or more and/or with a radial tank, the deltaV of my stage is broken.

This video explain the bug on the VAB, you can stop the video to read the text.

Now, the problem in my mission in exploration mode. I add 8 radial tank and 8 engine behind them. So my deltaV is broken. I have ~3600m/s for a ~1100m/s manoeuver but I use all my hydrogen before I finish my manoeuver.

image.png?ex=6596ee6f&is=6584796f&hm=7d5

 

 

Included Attachments:

Edited by The Space Peacock
Link to comment
Share on other sites

40 answers to this question

Recommended Posts

  • 5

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 10 | CPU: AMD Ryzen 7 2700X Eight-Core | GPU: NVIDIA GeForce RTX 2080 | RAM32GB DDR4 at 2933mhz

Specs:  - AMD Ryzen 7 2700X 3.7ghz, 8 cores / 16 logic cores, 32gb DDR4 running at 1466mhz/2933 effective, nVidia GeForce RTX 2080, video driver 31.0.15.4629, 8GB vram. No overclocking.

Severity: Low to Mid. (You can work around by trial and error, but this bug makes rocket design a MASSIVE pain)

Frequency: 100% reproduceable for me, with steps, including loading a craft file and having it exhibit the same wrong/inconsistent dV readings

 

Description:  I was making a launch vehicle when I noticed adding a relatively small cone tank DRASTICALLY changed the dV readout for the first stage, both in VAB and after launch. Flying the vehicle showed the dV coming up short of the readout in the staging. I made new simple test crafts in the VAB until I narrowed down a fairly minimum rocket that shows the wrong dV calculation in the VAB.

Steps for building craft:
-Go to VAB, start building a new craft from scratch
-Choose the large S3-14400 methalox tank as first part
-Add TD-37 stack decoupler on the bottom of the tank, with arrow up
-Add a second S3-14400 tank to the bottom of the TD-37 stack decoupler
-Add a KR-2XL Rhino engine to the bottom of the second tank
-Grab another S3-14400 tank from the part menu
-Switch radial symmetry mode to 2x, and radially attach that tank to the side of the second/lower tank on the vehicle
-Grab another KR-2XL Rhino engine and, using 2x radial symmetry, add it to the bottom of the radially-attached tanks
-Note the dV readout of 3,084 m/s
-Grab a small FL-T100 methalox tank from the part menu, and in 2x symmetry mode add it to the top of the radial S3-14400 tanks
-Note the dV jump up to 5,590 m/s
-Remove and add the FL-T100 tank and note the dV jumps between 3,084 and 5,590

This is far too large a dV jump for adding such a small tank to such a huge stage. When flying the original rocket it ran out of fuel at far less than the absurdly-high dV figure.
I don't know if it specifically requires the parts I listed in the steps so I just listed exactly what I was testing with, though it does seem to work when you add some other tank in place of the FL-T100. 

 

Included Attachments:

radialtankadddVbugrepro.json

Link to comment
Share on other sites

  • 5

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 11 | CPU: i7-11800H | GPU: RTX 3050 Laptop | RAM16GB

Frequency: High

Severity: Medium (misleading)

Δv calculation in a stage when there's only one engine is fine, along with TWR, Burn time and Start mass. Things get incorrect when there are multiple engines in a stage.

More engines make Δv tool make incorrect calculations to burn time and total Δv, TWR and Start mass are not effected.

Here are two screenshots of the same craft with 4 nervs and 1 nerv. 4 nervs must result in less Δv than 1 nerv but it is just wrong.

Included Attachments:

Link to comment
Share on other sites

  • 4

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 10 Home | CPU: AMD Ryzen 7 3700X 8-Core Processer | GPU: NVIDIA GeForce GTX 1650 Super | RAM32 GB Ram

 

If you stack a smaller fuel tank below a larger one, you get a smaller dV than if the smaller tank is above the larger one, when connected to an engine plate or adapter with two or more engines attached.  To reproduce the bug, it seems the set up must be part of a larger rocket (for instance with a decoupler above).  I also able to sometimes create other random dVs by attaching tanks radially with the same setup but have trouble reproducing (You may need a more complicated setup, like 2 tanks stacked serially and then others attached radially.)

  • x200-16 above x200-32: 3522 dV
  • x200-16 below x200-32: 1866 dV
    image.png?ex=65976735&is=6584f235&hm=522

 

 

 

Included Attachments:

Link to comment
Share on other sites

  • 4

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: windows 10 | CPU: 11th Gen Intel(R) Core(TM) i7-11370H @ 3.30GHz 3.30 GHz | GPU: NVIDIA GeForce RTX 3070 Laptop GPU | RAM16 gb

When adding a single engine to the bottom of this rocket the DV report says it should have 2238 m/s of dv, when putting two engines on the side of the rocket the dv goes up to , which it should not as the added engine should make the rocket heaveir and the dv should go down.

 

Included Attachments:

munrover.json

Link to comment
Share on other sites

  • 4
On 12/22/2023 at 2:20 AM, Anonymous Bug Reporter said:

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 10 Home | CPU: AMD Ryzen 7 3700X 8-Core Processer | GPU: NVIDIA GeForce GTX 1650 Super | RAM32 GB Ram

 

If you stack a smaller fuel tank below a larger one, you get a smaller dV than if the smaller tank is above the larger one, when connected to an engine plate or adapter with two or more engines attached.  To reproduce the bug, it seems the set up must be part of a larger rocket (for instance with a decoupler above).  I also able to sometimes create other random dVs by attaching tanks radially with the same setup but have trouble reproducing (You may need a more complicated setup, like 2 tanks stacked serially and then others attached radially.)

  • x200-16 above x200-32: 3522 dV
  • x200-16 below x200-32: 1866 dV
    image.png?ex=65976735&is=6584f235&hm=522

 

 

 

Included Attachments:

And to be clear, this IS NOT just a VAB UI bug like the one that causes all stages to show 0dV in the VAB, but shows values in flight.

THIS issues actually persists into flight.

Link to comment
Share on other sites

  • 4

I can confirm this bug - it is terrible. It messed up my Tylo lander with 2 Thud engines and some spherical tanks attached, indicating ~3000 m/s when in fact it had ~1600 m/s. Jeb got stranded on Tylo... and he's mad!

Link to comment
Share on other sites

  • 4
On 12/29/2023 at 3:05 AM, Matrix Aran said:

The only thing I have to add to what has already been said is that in one test case, I had a pair of NERV engines attached to radial attachment nodes. Every time I cloned the nodes and placed them nearby, the DV calculation would increase. This was to try to work around clustering on an engine plate. 

To give more details about this case:

This craft has a stage with hydrogen to be used by the NERV, then on top a decoupler and finally on top of it a methalox tank.

First picture, the methalox tank is empty, and the Δv calculation is correct
 

Spoiler

XOz3VUX.jpeg

Second picture, I fill methane in the methalox tank, and the Δv displayed for the NERV stage increase
 

Spoiler

tnqrrS1.jpeg

Third picture, I fill the oxidizer in the methalox tank, and the Δv displaye for the NERV stage increase further
 

Spoiler

PFO36MI.jpeg

When calculating the Δv, the methalox tank is treated as a source of fuel for the NERV engines.
You have to remember that:

  • Methane and oxidizer are not supposed to be used by the NERV engine
  • There is a decoupler with fuel crossfeed disabled between the methalox tank and the hydrogen tanks.

The craft used for this test is below available there: https://pastebin.com/mTzh30sY 

Link to comment
Share on other sites

  • 3

Anonymous Bug Reporter : I can reproduce this bug too. Same problem for the hydrogen tank.

 

The Space Peacock : I tried to use engine mount or adapter but the bug is still there.

Without : 5022m/s

With engine mount : 5002m/s

With adapter : 5013m/s

The deltaV is moving just because of the weight of the adapter/engine  mount.

KSP2-x64-n9-S8t-WWWe-A.jpg

Link to comment
Share on other sites

  • 2

Has anyone tried replicating these results with the rocket equation to narrow down where the error might be? It's clearly not just leaving out the additional mass, because that would give the same delta V, not greater. Maybe it (also) thinks the exhaust velocity increases with the increasing thrust?

Link to comment
Share on other sites

  • 2

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 10 | CPU: AMD Ryzen 5 2600 Six-Core Processor | GPU: RTX 2060 | RAM16

I've been consistently noticing an issue with the delta v readouts (espcially in For Science) where it will tell me that my craft has 1.5-2 times more delta v than it actually has. This has been a really big issue for me as it has made craft with more than one stage often impossible to fly.  I've noticed this bug since launch, but its been especially noticeable in For Science.

Included Attachments:

fordevs.mp4

wwe.json

Link to comment
Share on other sites

  • 2

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 11 | CPU: Ryzen 7 3700X | GPU: RTX 3070 | RAM32

While constructing a rocket adding an additional fuel tanks in 2-times symmetry modified the deltaV way more than it should. Switching from vacuum to atmosphere or celestial body does not fix it.

Adding only 1 does seem to change the deltaV in a correct amounts, but after adding the second the error happens again.

Part Added (2 times symmetry) Change in deltaV
FL-T800 439
FL-T400 1613
FL-T200 1993
FL-T100 1949

Adding the tanks to the "central one" doesn't cause the bug

Included Attachments:

deltaVerror.mp4

MunLander.json

Link to comment
Share on other sites

  • 1

I have same kind of problem here. i have 90t vessel, if i have 1 swivel it shows 2621dv.

Adding engines gives weird values:

1swivel=2621

2 swivel=1570

3swivel=1539

4swivel=1510

so it shows that dv calculations are correct when there is only 1 engine, with some adapters it seems to give wrong dv. Same seems to happen with engine mounts.

Edited by Jeq
Link to comment
Share on other sites

  • 1

The only thing I have to add to what has already been said is that in one test case, I had a pair of NERV engines attached to radial attachment nodes. Every time I cloned the nodes and placed them nearby, the DV calculation would increase. This was to try to work around clustering on an engine plate. 

Link to comment
Share on other sites

  • 1

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 11 Home | CPU: i7-13700k | GPU: GTX 1080 Ti | RAM32GB

 

Context: Testing Jool encounter / low orbit stage (teleported it into LKO) | Craft starts with 5,700~ delta-v in LKO. Getting an intercept with Jool takes about 2,400~ delta-v with the node I set; thus we have 3,300~ delta-v to work with. The burn to intercept Jool was time warped quite high, and did not have any issues. I've had this problem occur with Duna, multiple times, and was forced to use Terrier engines in order to orbit Duna.

 

This video was recorded after a time-warped burn that completely failed, it got less near orbit than the full 5+ minute burn. The recording shows the first 5+ minute burn in full, after it fails I try time-warp again.

Also, it displays a weird bug where I'm forced to exit and re-enter map view to edit my node, the manuveur node on the navball getting reversed after quicksave load (many times encountered) and also delta-v calcuations heavily impacted just by transferring fuel.

Notes:

I have never even tried to build a Jool capable stage before. In KSP 1 & 2, I've only gotten to Duna & Ike / Kerbin's moons. This is intended to stay in Jool orbit. So it's probably really badly designed. However, I'm positive it's failing due to buggy behavior.

*As on 1/14/2023, 5:01pm, the video is estimated to process to 1080p+ in one hour

Included Attachments:

Link to comment
Share on other sites

  • 1

Reported Version: v0.2.0 (latest) | Mods: none | Can replicate without mods? Yes 
OS: Windows 11 Home | CPU: i7-13700k | GPU: GTX 1080 Ti | RAM32GB

The below screenshot shows what occurs when the fuel tanks are sharing fuel; as you can see, the delta-v is showing as 3,982. However,

2fehEFi.jpg

 

Once you make it to where there is no crossfeed; suddenly there is 5,548 delta-v. The only change here, was disabling Fuel Crossfeed. It calculates nearly 1,600 more/less Delta-V, only because of fuel crossfeed.

FM72hJ8.jpg

This is leading to misleading/outright incorrect knowledge of how much fuel you really have. I encounter this consistently when attempting to build something with NERV engines especially. It doesn't seem to be so wildly off with methalox engines.

Included Attachments:

Link to comment
Share on other sites

  • 1

I played around with several launch system designs and had several launches hit by these bugs. The best I can tell, there are at least two issues at play. Radial tanks and engine plates with multiple engines can both cause inflated dV readouts. These are not by small amounts either. I saw multiplicative errors with overestimated dV by 300%. That is, rockets with 4k dV listed running out of fuel at 30km launching from Kerbin.

It makes planning a launch craft design very difficult as the only way to trust design results is to run a full launch test. Sometimes you can play games removing and reattaching engines to try to get it to recalculate the correct dV, even if only temporarily before returning the extra engines or radial mount tanks for the final assembly.

I was able to reproduce some of the cases described earlier such as radial mounting the smallest methalox tank to have it double the dV on an XL rocket. Clearly a faulty readout. I've upvoted the repro cases above reflecting scenarios similar to my cases.

Link to comment
Share on other sites

  • 1

I have been able to consistently reproduce a dv calculation error in the VAB, using the following steps:

KSP version shows as:    0.2.1.0.30833

1) clear workspace

2) top stage:   HECS2 pod / FLT200 fuel / LV909 engine / TD-12 decoupler

3) mid stage:  RS-AD 800 fuel / X200-16 fuel / RE-L10 engine / TD-25 decoupler

4) lower stage:  ADTP-2-3 fuel / S3-7200 fuel

5) add a single S3-KS25 "vector" -- stage 1 dv calcs:  ASL 2749 / VAC3038      (total dvs:  6298/9970)

6) remove the vector engine.   place it "radially" to the bottom in 2x symmetry (or more) so there are now 2 vector engines.   -- stage 1 dv calcs:    ASL 4207 / VAC 4650     (tot dvs:  7757/ 11582)

 

-- As you can see, the dv calculation was higher with 2 engines, than a single engine, which cannot be correct.  Not sure if the single engine dv was wrong, or the 2 engine dv was wrong.

 

Notes:

When the mid stage or the bottom stage have only 1 fuel tank, I dont seem to be able reproduce it.    

With less than 3 stages, I havent been able to reproduce it.

So this may be the simplest example.

 

I hope this helps find the issue!      Good Luck and thanks for all the hard work!

-cft2

 

Edited by CFTeague2
minor clarification
Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...