Rocket dispersion - research and tests

[Avimimus]

Hello all,

Cards on the table: I think gameplay would be improved if rocket dispersion was higher. It would make the rockets more useful - especially in automatic mode / salvo use. I also tend to suspect that the numbers given for rocket dispersion by manufacturers or airforces tend to be a bit optimistic, and that at least a small percentage of rockets fired would have much higher dispersion than the others.

However, I lack enough data to be confident I'm right about this - and I don't want to bother the devs about this unless I have evidence that dispersion should be higher.

The HVAR has a reputation for being unusually accurate:

Historical documents often give 5 angular mils as the dispersion of the HVAR (presumably under ideal conditions), although some sources list 7 mils or 10 mils (possibly closer to what was achieved in service).

Testing dispersion in game:

I created a mission with a parked P-51 and a triangular lattice (25 metre spacing) of Ba-64 armoured cars at 5.5km. This allowed firing rockets repeatedly at a consistent range:

HVARDispersion.zip

From this I can draw a few conclusions:

- If my math is correct, at 5500 metres 5 angular mils should lead to an area 100 metres wide. This is consistent with my tests.

- The rocket patterns seem pretty random and there seems to be less of an issue of the rockets forming a torus (or donut) as was the case with the R4M in Great Battles. This suggests the team may have improved the algorithm.

- The rocket impacts appear to have more dispersion in range than in lateral arc. I suspect that this is largely due to the low angle of fire (leading to the projection of a circular dispersion pattern onto an angled surface, which appears as an oval). However, it is quite possible that the team has modelled some random variations in rocket engine power (which would be realistic). Viewing the rockets in flight seems to show some rockets flying slightly faster than others, which would support this.

- I have an impression that later salvos fired have higher dispersion than the first salvo fired. However, I'm uncertain whether this is a bug or whether this is just a matter of probabilities (more rockets fired means that rockets that have more chances to get off course).

 

Misgivings:

- I'm wondering if 5 mils might be CEP50 whereas the sim seems to treat 5 mils as CEP90? In other words, 90 percent of HVAR seem to be landing within the 5 mil radius, whereas the historical documents might be stating that 50 percent would land within 5 mils and 50 percent would land outside of 5 mils?

- A CEP50 of 5 mils might be what was achieved under ideal conditions (with 7-10 mils more common in practice)? That could explain why historical documents and manuals differ for the precision of the 5" HVAR. I also wonder if turbulence from the initial missiles might have impact the accuracy of later missiles in the same salvo. I'll admit to having some trouble imagining unguided rockets being this accurate.

- The tests are slightly complicated by the fact that the HVAR warhead seems to be able to destroy a Ba-64 at 15-20 metres (and the spacing of the lattice may have a degree of error as the vehicles aren't always exactly 25 metres apart).

Anyway, feel free to give this a try and let me know what you think. In addition to static firing, one can take off. It is also useful to try this mission with unlimited ammo turned on (good for practice).

I'm quite impressed - with my main concern might be the possibility of confusing the CEP50 with CEP90!

[AEthelraedUnraed]

Interesting research!

One should not forget that in combat, you'll always get worse results than when testing, simply because aiming isn't an issue in the latter case. I'm lucky if I can get any rocket at all within 50m of the target, let alone at a range of 5500m😅

39 minutes ago, Avimimus said:

- If my math is correct, at 5500 metres 5 angular mils should lead to an area 100 metres wide. This is consistent with my tests.

I think that should be 50m wide:

5500*tan(5/1000 rad)=27.5m per side, so 55m total. 100m corresponds roughly to 9 mils. That would nicely correspond to your stated figure of 7-10 mils.

[Avimimus]

6 minutes ago, AEthelraedUnraed said:

Interesting research!

One should not forget that in combat, you'll always get worse results than when testing, simply because aiming isn't an issue in the latter case. I'm lucky if I can get any rocket at all within 50m of the target, let alone at a range of 5500m😅

I think that should be 50m wide:

5500*tan(5/1000 rad)=27.5m per side, so 55m total. 100m corresponds roughly to 9 mils. That would nicely correspond to your stated figure of 7-10 mils.

Thanks for correcting the math - I'm definitely not used to calculating angular mils.

Note: There is the added effect that the blast radius will destroy a vehicle at 15-20 metres... so the actual destruction needs to have about 30-40 metres removed. So, in retrospect, my tests probably reveal a dispersion of something more along the lines of 60-70 metres.

Anyway - download the mission and give it a test. The aircraft is already lined up to act as rocket artillery firing from the runway - and one can use unlimited ammo to fire off multiple salvos.

[Jaegermeister]

Well, if you are comparing in game to real world results during the Korean War you would have to take another factor into consideration. From stories I have read, all of the HVAR rockets used in the early part of the Korean War were left over from WWII. They had been stored on an Island in the Pacific and stacked for a few years in warehouses. The propellant settled to one side, and solidified due to the high heat and humidity. This caused unbalanced weighting which varied according to how they were stacked, and resulted in unpredictable results when they were loaded on the wing racks. The pilots stated that they went everywhere except where they were aimed. If this is true, the in game results will almost surely be much more accurate than they were in 1950 and 1951. What I have seen in game so far is nothing near as bad as what is described in anecdotal stories from the time.

Not to take anything away from your detailed testing, I just wanted to throw that out there.

 

Edited July 7 by Jaegermeister

[Dash,Polder]

You negate everything with unlimited ammo on, because balance, drag, and weight no longer affect the aircraft's dynamic inputs on the rockets.  Don't believe it, than take a Mustang with two napalm canisters, drop one and fly around a bit, than drop all and reset the weight, now she flys perfectly straight no matter what or how you drop from that point.  Same with rockets, all testing after initial firing of the complete first load is invalid, because the physics dynamics no longer reflect a real modeling state.  It's good enough to help you practice and familiarize with the weapons, but not make accurate decisions on the modeling of them.

I think the fin stabilization physics simply blows the house away, all you have to do is apply full rudder in level flight and loose a rocket to see the dynamics in action.  Totally nailed it, they have an equal and opposite reaction to the slip and drag induced on their fins and go sidewinder in the wobble.  Never saw that in any other sim.  I love the job they've done here.  I can feel the disturbance in balance on a Mustang once that first rocket departs the left wing rail, more pronounced on the shaped charge anti-armor rockets with the larger diameter warhead.  But you don't get that anymore once your on a reloaded wing with unlimited weapons.   

[Avimimus]

1 minute ago, Dash,Polder said:

You negate everything with unlimited ammo on, because balance, drag, and weight no longer affect the aircraft's dynamic inputs on the rockets.  Don't believe it, than take a Mustang with two napalm canisters, drop one and fly around a bit, than drop all and reset the weight, now she flys perfectly straight no matter what or how you drop from that point.  Same with rockets, all testing after initial firing of the complete first load is invalid, because the physics dynamics no longer reflect a real modeling state.  It's good enough to help you practice and familiarize with the weapons, but not make accurate decisions on the modeling of them.

I think the fin stabilization physics simply blows the house away, all you have to do is apply full rudder in level flight and loose a rocket to see the dynamics in action.  Totally nailed it, they have an equal and opposite reaction to the slip and drag induced on their fins and go sidewinder in the wobble.  Never saw that in any other sim.  I love the job they've done here.  I can feel the disturbance in balance on a Mustang once that first rocket departs the left wing rail, more pronounced on the shaped charge anti-armor rockets with the larger diameter warhead.  But you don't get that anymore once your on a reloaded wing with unlimited weapons.   

Ah, yes. But the angular mils quoted are probably for the rocket itself as fired under relatively ideal conditions - so firing from the runway with unlimited ammunition turned on is a lot like firing from a ground based test stand.

[AEthelraedUnraed]

44 minutes ago, Avimimus said:

Ah, yes. But the angular mils quoted are probably for the rocket itself as fired under relatively ideal conditions - so firing from the runway with unlimited ammunition turned on is a lot like firing from a ground based test stand.

I'm pretty certain. I don't believe in the slightest that the average pilot would be able to get a HVAR within 50m of the target at a distance of 5.5km.

[Avimimus]

46 minutes ago, AEthelraedUnraed said:

I'm pretty certain. I don't believe in the slightest that the average pilot would be able to get a HVAR within 50m of the target at a distance of 5.5km.

If everyone would read the post carefully - there is a link to a zip file which has a custom mission. In the custom mission there is a P-51D idling on the runway. If the pilot fires the rockets they will fly 5.5 kilometres (katyushka style) and land in a field of Ba-64 armoured cars which I have positioned. The field is hundreds of metres wide.

This was done to test the ballistic accuracy of the HVAR as modelled in the sim when fired at sea level.

[Uther]

2 hours ago, Jaegermeister said:

Well, if you are comparing in game to real world results during the Korean War you would have to take another factor into consideration. From stories I have read, all of the HVAR rockets used in the early part of the Korean War were left over from WWII. They had been stored on an Island in the Pacific and stacked for a few years in warehouses. The propellant settled to one side, and solidified due to the high heat and humidity. This caused unbalanced weighting which varied according to how they were stacked, and resulted in unpredictable results when they were loaded on the wing racks. The pilots stated that they went everywhere except where they were aimed. If this is true, the in game results will almost surely be much more accurate than they were in 1950 and 1951. What I have seen in game so far is nothing near as bad as what is described in anecdotal stories from the time.

Not to take anything away from your detailed testing, I just wanted to throw that out there.

 

In my experience with the rockets I am reenacting this bad storage perfectly because when I fire, with my aim they do indeed go everywhere except the target! 😂

[Dash,Polder]

1 hour ago, Avimimus said:

Ah, yes. But the angular mils quoted are probably for the rocket itself as fired under relatively ideal conditions - so firing from the runway with unlimited ammunition turned on is a lot like firing from a ground based test stand.

And your own results seem to imply there indeed isn't the same weight on the following volleys. 

 

4 hours ago, Avimimus said:

- I have an impression that later salvos fired have higher dispersion than the first salvo fired. However, I'm uncertain whether this is a bug or whether this is just a matter of probabilities (more rockets fired means that rockets that have more chances to get off course).

Why you can't use unlimited weapons to test, it isn't using the same physics model anymore.

[AEthelraedUnraed]

2 hours ago, Avimimus said:

If everyone would read the post carefully - there is a link to a zip file which has a custom mission. In the custom mission there is a P-51D idling on the runway. If the pilot fires the rockets they will fly 5.5 kilometres (katyushka style) and land in a field of Ba-64 armoured cars which I have positioned. The field is hundreds of metres wide.

This was done to test the ballistic accuracy of the HVAR as modelled in the sim when fired at sea level.

I understand 🙂 That's why in the post you quote, I agree with you that the numbers you quote are "for the rocket itself as fired under relatively ideal conditions" 😉

 

[Avimimus]

Yeah, sorry - I feel like I'm losing it here... I'm really overthinking this - and we have some figures for dispersion (although I no longer seem to recall whether they are CEP50 or CEP90) and we don't really know what the dispersion was around that average dispersion... if that makes sense?

[Gunfreak]

I haven't done much scientific testing. 

But a shallow dive trim the aircraft stable. And it's all down to judging the drop. Unless you after blancketing area. They're relatively close range weapons. 

Still haven't gotten a perfect hit on armor yet. But I've had a few rockets miss by just a meter. 

[ArgonV]

I can't hit squat with the rockets! I was pretty decent in GB, so I suspect the physics modelling in Korea WRT rockets is more realistic.

[Avimimus]

13 minutes ago, ArgonV said:

I can't hit squat with the rockets! I was pretty decent in GB, so I suspect the physics modelling in Korea WRT rockets is more realistic.

I do get that impression - there seems to be some variation in the rocket motor burn times (a real issue with solid fuel rockets), and I suspect dispersion is less likely to produce a torus (which seems to be an issue with the R4m etc. in Great Battles).

That is the thing really: They've done a really good job overall - which is why I suspect that my intuitions are wrong.

That said, I still find it hard to believe that a fin stabilised rocket can have this much ballistic accuracy (especially when launched from a stand-still)!

8 hours ago, Gunfreak said:

I haven't done much scientific testing. 

But a shallow dive trim the aircraft stable. And it's all down to judging the drop. Unless you after blancketing area. They're relatively close range weapons. 

Still haven't gotten a perfect hit on armor yet. But I've had a few rockets miss by just a meter. 

I spent quite a bit of time looking for gun camera footage from the Korean War - and most of the footage shows single rockets being launched at area targets. Essentially firing, observing the fall of shot - and then adjusting - but also gradually switching targets.

[Dash,Polder]

Perfect trim with no side slip, fly long circuits and start an approach like four miles at 4,000 feet, keep your dive angle consistent.   Each firing adjust the deflection on your sight till your on target.  No rudder correction on final approach.  I use the + only on target with the rocket gyro on in Mustang, but for slip indication only.  They can be hit, just takes some precision flying.

[Avimimus]

To my considerable amusement I discovered a thread that sounded like it might have useful data, then noticed who the originally created the thread twenty years ago:

https://www.key.aero/forum/modern-military-aviation/missiles-and-munitions/61546-rocket-dispersion-and-accuracy-refs-and-pictures

 

 

NOLTR 63-110 is interesting: https://apps.dtic.mil/sti/tr/pdf/AD0474776.pdf

1. It includes a nice summary of equations governing the dispersion of an unguided air-to-ground rocket. It makes a very interesting technical read.

2. It implies that the shift to spin-stabilised folding fin rockets was mainly to allow cheaper production, logistics, and stowage (e.g. use in rocket pods). This is consistent with the fact that these later rockets often had similar or greater dispersion to the HVAR. There may be an implication here that manufacturing defects or damage in handling is more of an issue for fin stabilised rockets.

3. It emphasizes that the biggest factor in inaccuracy is off-centre thrust from the solid fuel rocket motor. The main benefit from spin stabilisation is not the gyroscopic forces but the fact that it averages out thrust asymmetries! Since the HVAR does not rotate in flight, the main source of inaccuracy is probably asymmetric burning of the motor. @Jaegermeister's contribution is particularly relevant in this case.

4. It mentions a good pilot has about '2-3 mil accuracy' in selecting an aiming point. This suggests that a salvo of rockets with a dispersion of more than 3 mil should reliably blanket a target.

 

OSRD-2545 on rocket fuses: https://www.bulletpicker.com/pdf/OSRD-2545.pdf

- Table 1 gives the burn time and acceleration of each rocket (HVAR/AR/11.75") depending on ambient air temperature 

- page 172 has charts from test firings of Mk.1 HVAR rockets showing their output at three different temperatures across their entire burn time

- information on fusing and underwater performance (as well as preferred doctrine for use in anti-shipping strikes e.g. p.154) as well as data on anti-fortification penetration (e.g. p175).

 

NAVWEPS OP2210 is a much later document (1960) but still has a couple of interesting references:

- 5-11 (i.e. chapter 5 page 11) has C.G.  locations for the HVAR

- It also references USN trajectory table OP No.1829 (but I've been unable to locate a copy)

 

NDL4009801, FF12-5/A8, SBB-15-Jes, Ser: 04139. 1945 report on use of 5" air-to-ground rockets https://upload.wikimedia.org/wikipedia/commons/d/d6/NDL4009801_Selection_of_rockets_for_destruction_of_various_targets%2C_June_1945._Report_No._11-a(19)%2C_USSBS_Index_Section_6.pdf:

- CEP50 of 12 metres under combat conditions

- It mentions that RAF Coastal Command reports that solid rocket warheads (i.e. AP without a bursting charge) are twice as likely to sink a ship. This suggests that modelling of flooding is important for ship damage models (and we could do with a boost to AP rounds against ships in Great Battles) 😉

- The 5" AR with Mk.1 mod.1 equipped with Mk.172VT fuse is also mentioned. I am not sure if this was used alongside the AR rockets in the Korean theatre? It is recommended for suppressing anti-aircraft guns and attacking personnel and vehicles.

- Interesting discussion of fragmentation, and limitations of SAP warheads for penetration against armour due to fragility (but assessing them as effective against concrete if properly fused), and penetration against different types of structures.

- Doctrine on appropriate targets for each type of rocket (useful for mission makers - and potentially useful for the campaign generators as well):

Spoiler

- Delayed fuses should be used against unarmoured warships, merchant vessels (excluding small craft), buildings, concrete installations, locomotives, and large fuel tanks. Otherwise instantaneous fragmentation fuses should be used. Against runways, bridges, railroad tracks, and docks bombs should be used (rather than delayed fused rockets)- however the 11.75" rocket can be used aginst bridges, heavy guns, tunnels, and railroad fills.
- Solid AP (3.5") rockets without bursting charges are preferred against destroyers, submarines, merchant vessels, and minor naval targets due to their greater ability to travel underwater and produce hits below the waterline, with 1" penetration at 50-60" of underwater travel reported (compared to a delayed fuse 5" rocket).  Shallow attacks should be used to avoid the rockets travelling under the ship. They can also be used to damage the internal compartments of ships when fired from above (although HE warheads will be slightly more effetive for any targets within their reach). Use of AP rockets against ships is assessed by RAF Coastal Command as being twice as likely to think the ship. 
- pp8-10 have tables providing doctrine for employment of weapons against various targets.

Finally, it includes a table on the lateral dispersion of 5" HVAR rounds under combat conditions in 1945. Launch range is not given. I have reproduced it below (converted to metres radius and percentages). Methods are not given in the document but might be derived from actual combat assessments - which could explain the lack of data regarding the launching aircraft's speed/firing range:

Note: The datapoints in this chart look a bit overly tidy - which suggests a certain amount of interpolation (and likely fitting a normal distribution), but it is still remarkable in being an attempt in wartime to give a more realistic assessment of accuracy in actual combat conditions.

It contrasts significantly with the 2-3 mil pilot error and 5 mil (or 7-10 mil) dispersion given for the HVAR in documents based on estimates produced under more ideal conditions (unless the rockets being fired under combat conditions were being fired at extreme range).