I've long wanted to do an analysis of how bomb accuracy has changed over the last 80 years or so, and how this impacts the cost of destroying a target from the air. The problem has always been that all of the data I need, from bomb accuracy to the cost of flying the planes, is scattered and sparse, and so I haven't really made much progress. But a recent comment from cassander inspired me to give it a try in Command: Modern Operations, which is basically a medium-level simulation package that is sold on Steam. I've used it for other experiments, and it could give at least a fairly reasonable number for how many bombs it takes from a given platform/delivery mechanism to destroy a generic target.
The basic setup I used for all of these tests was fairly simple. I set up a grid of 10 medium buildings (130' x 80') in CMO and then on the opposite side set up a nearby airbase with 12 planes of the type under test. All were given the same loadout (usually the heaviest one for the weapon in question) and set up to attack one at a time, using only one bomb at a time on a given target to prevent overkill from messing up my numbers. I would run the scenario until either all of the targets were destroyed or all of the planes had flown their sorties. Variables included the plane, the weapon, attack altitude and environmental conditions (day/night and weather).
For my first test, I ran a simple scenario: a modern fighter (Super Hornet) using dumb bombs, able to attack as low as it liked, with no weather conditions or other problems. The results were pretty straightforward:
| Plane | F/A-18E |
| Weapon | 8 x Mk 82 |
| Conditions | Clear Day |
| Altitude | 875' |
| Total Bombs | 27 |
| Hit % | 100% |
| Avg. Miss | 29' |
| Results | All destroyed |
So at this point, it's probably worth talking a bit about methodology. I'm going to count as a hit any bomb which does damage to the target, which can include both direct hits (16 of the 27 here, and, yes, many buildings survived direct hits) and bombs that don't hit directly but still do damage. Average miss counts the distance as 0 for direct hits, which is obviously somewhat optimistic (the target does have dimensions, so add maybe 25-30' to get actual CEP). For the next run, let's use the same loadout, but raise the altitude to 12,000', the sort of altitude you'd use if you wanted to avoid being shot at with MANPADS and guys with rifles.
| Plane | F/A-18E |
| Weapon | 8 x Mk 82 |
| Conditions | Clear Day |
| Altitude | 12000' |
| Total Bombs | 67 |
| Hit % | 66% |
| Avg. Miss | 211' |
| Results | All destroyed |
Obviously, performance this time was significantly worse, with a lot of the bombs falling so far wide that they did no damage, although there were also 12 direct hits on the 10 buildings, which strongly suggests that the direct hits do the vast majority of the damage, an impression confirmed by later tests. That said, this one took long enough that several of the buildings were only damaged by the bombs and ultimately destroyed by fire. But the obvious next question is how this would work with bigger bombs, so I replaced the Mk 82s (500 lb) with Mk 83s (1000 lb) and ran the scenario again.
| Plane | F/A-18E |
| Weapon | 8 x Mk 83 |
| Conditions | Clear Day |
| Altitude | 12000' |
| Total Bombs | 38 |
| Hit % | 74% |
| Avg. Miss | 202' |
| Results | All destroyed |
We see slightly better performance here, mostly due to the greater lethal radius of the bombs. In terms of total tonnage, it took slightly more Mk 83s, but fewer sorties because there were more bombs per plane. But it's also worth going up the final step and seeing what happens if we use 2000 lb Mk 84s.
| Plane | F/A-18E |
| Weapon | 4 x Mk 84 |
| Conditions | Clear Day |
| Altitude | 12000' |
| Total Bombs | 22 |
| Hit % | 74% |
| Avg. Miss | 190' |
| Results | All destroyed |
This was the clear winner in terms of number of bombs required, although relative to the Mk 83, 6 sorties were required rather than 5 because the plane could only carry half of the bomb load. But with that baseline out of the way, I think I'm going to go with the Mk 83 for further tests with factors that could affect unguided bombs. First, I'll get a low-altitude baseline for comparison in various conditions:
| Plane | F/A-18E |
| Weapon | 8 x Mk 83 |
| Conditions | Clear Day |
| Altitude | 1100' |
| Total Bombs | 14 |
| Hit % | 93% |
| Avg. Miss | 28' |
| Results | All destroyed |
It's worth noting that the low-level attack with Mk 83s required only about half the bombs that the Mk 82 version did, largely because the bombs were far more likely to destroy the building in a single hit. This definitely confirms the decision to use the Mk 83 as my standard test bomb when possible. Next, let's see what effect night has on the attack.
| Plane | F/A-18E |
| Weapon | 8 x Mk 83 |
| Conditions | Clear Night |
| Altitude | 1100' |
| Total Bombs | 14 |
| Hit % | 100% |
| Avg. Miss | 36' |
| Results | All destroyed |
This test showed very slightly worse performance in terms of miss distance than the low-level day attack (hit percentage was higher, but the only "miss" in the day test was a bomb that malfunctioned). But it's not really that surprising that a fully kitted-out Super Hornet, with ATFLIR and ASG-34 IRST can deliver weapons nearly as effectively in the dark. Let's see how well that holds up to going to medium altitude.
| Plane | F/A-18E |
| Weapon | 8 x Mk 83 |
| Conditions | Clear Night |
| Altitude | 12000' |
| Total Bombs | 42 |
| Hit % | 90% |
| Avg. Miss | 198' |
| Results | All destroyed |
The medium altitude results are quite similar to the daytime ones in terms of average miss distance and total bombs required. Hit percentage was substantially higher, although I think this is probably not the best metric to evaluate these tests by. It counts any bomb which does any damage, so it's easy enough for a lot of bombs landing 250'+ out, which really aren't doing very much to the building, to count as hits.
But now that we have a modern baseline, it's time to start looking back. I'm going to start with one of the earliest multirole fighters in this database,1 the F-4J Phantom II, largely because it can carry the same Mk 83 bombs that I've been using here. We'll start at low altitude.
| Plane | F-4J |
| Weapon | 3 x Mk 83 |
| Conditions | Clear Day |
| Altitude | 1100' |
| Total Bombs | 26 |
| Hit % | 100% |
| Avg. Miss | 140' |
| Results | All destroyed |
Here, we see a significant falloff in ability to do damage. Most of the bombs are still getting close enough to do damage, but it's gone from 14 to 26 to wipe out all the buildings. I'm going to try taking this to altitude next, although there's an obvious problem waiting. The Phantom's bombload is only 3 bombs, and it took 38 for the Super Hornets to destroy all the buildings from medium altitude. I'm going to set each of the Phantoms up to fly twice, to ensure a reasonable number of bombs get used.
| Plane | F-4J |
| Weapon | 3 x Mk 83 |
| Conditions | Clear Day |
| Altitude | 12000' |
| Total Bombs | 72 |
| Hit % | 33% |
| Avg. Miss | 648' |
| Results | 8 destroyed, 2 heavily damaged |
This was obviously a massively different result than the Hornets from medium altitude, and a good indication of why the arrival of PGMs was so revolutionary. At the same time, the Phantom was not exactly known for being one of the best bombing platforms of the day, so let's try a different plane, the 1975 version of the A-6 Intruder, again attacking from medium altitude.2
| Plane | A-6E |
| Weapon | 10 x Mk 83 |
| Conditions | Clear Day |
| Altitude | 12000' |
| Total Bombs | 74 |
| Hit % | 59% |
| Avg. Miss | 401' |
| Results | All destroyed |
This was an interesting comparison to the Phantom. The average performance was significantly better, with a lot fewer genuine flyers landing 1000'+ from the target, but it wasn't all that much better at actually landing the sort of hits that counted. Yes, it got all buildings in 74 bombs instead of 8/10 in 72, but most of the "hits" were 200'+ away and didn't do all that much damage.
I ran a night test with the Phantoms, and the results were extremely similar to the day test, at least until I tried to send out the second wave and it wouldn't launch, because that bombing mission was day-only. I'm not sure how I got the first flight off, or how the game models night in general, but I think this post has run its course. I'll pick up the subject later.
Also, on an unrelated note, CMO has just rolled out multiplayer for the standard game, so if anyone is interested in contributing to future posts like this by letting me put humans on multiple sides, let me know. To be clear, this is less "let's see who is better" and more research like in this post. But I'm curious to see what I can do in cases where I'm less limited by my ability to script things.
Comments
...Well done, sir, well done. I would very much like to help out with any further tests, but it wouldn't be until February, when I retire and will have plenty of time to assist.
And BTW, have you seen anything about Project HANNIBAL - basically it looks like they're unleashing the ground combat systems.
Mike
This is interesting, but I do wonder how much of the result is “the observed performance by NAVAIR during testing” vs “differences in how the game encodes and interprets those parameters”? Defense games tend to have a lot of dedicated enthusiasts to ensure the data is right, but it still seems like a decent amount is still “this is what the game engine thinks”.
Cool stuff. I have the same concern redRover has - this is a model, not actual conditions, so it's useful, but not absolutely predictive. It would be interesting to see just how much CMO tries to take into account.
The big thing here is that you haven't added AA to the base, which we naturally would expect to complicate things mightily.
A simpler factor, though, would be the terrain. I'm assuming flat ground in your tests. I would expect different results if the facilities are on a slope, and depending on whether the bombers run upslope, down, or cross.
@redRover
Yes, that is an issue, and I don't have a good answer beyond "I trust their model more than anything else I can find to generate these kind of answers". I've looked for this kind of stuff before several times, and largely came up blank. At least there actually is a model at work here, and not just some performance figures found in a random document.
@Paul
I would say that the AA is represented by going from minimum altitude to 12,000'. Part of this is that I'm trying to isolate my variables, and adding AA is exactly the opposite of that, because the results start to get dependent not only on bomb accuracy, but also on what gets shot down.
Hadn't thought about terrain, may try that at some point.
I agree with that reasoning for bypassing AA, yeah.
Depending on what your slope experiments reveal, I notice there could be benefit in real world scenarios in using the terrain to force a bombing approach that makes AA more effective. Makes me wonder how DoD approaches this.
How close are the buildings, relative to their 130'x80' footprint (and the average miss distances)? With 60+ bombs per run, I imagine there would have been several cases of a bomb targeted at one building scoring a near miss or indeed direct hit on another building. Or to operationalize it: if you take the number of bombs (sorties) required to demolish the target in this manner, but set them up to just carpet the target area in one pass, what fraction of the buildings do they destroy (damage)?
JDAM already has so many options, the hollow tailcone could be used to house a BDA submunition. A webcam on a parachute, more or less. (I assume that in the game, the planes can know immediately and with 100% reliability that a target has been destroyed, and so can choose mid-sortie what target to service on their next pass. Otherwise, if the dust and smoke are modeled, then there should be significant differences between the first and subsequent sorties' accuracy.)
I pulled the file. The test was indeed conducted in a flat area, and I haven't gotten around to doing a slope test yet. (That would be harder, because I couldn't just let the AI do all the flying.)
Re buildings, I was aware this was a potential problem, and the closest pair are just over a mile apart. I don't recall ever seeing a bomb intended for one building damage another. Obviously, this isn't how it would work in real life, but I'm trying to isolate variables here. And no, I don't think smoke is modeled either.
BDA in CMO isn't totally perfect, and I would often get updates when a plane overhead spotted a fire or something, but you do learn instantly of buildings being destroyed.
I put the question 'The cost of destroying a target from the air' into Grok AI and got the below:
"I’ll look up publicly reported costs of air-delivered munitions and typical strike packages so the answer is grounded in current figures, not guesswork.
There is no single price. In state military operations, the bill is the munition plus the aircraft (or drone) that delivers it, and those two pieces can differ by two or three orders of magnitude.
Munition only (public U.S. procurement-style unit prices, roughly):
Class Typical weapon Ballpark unit cost Cheap precision bomb JDAM kit on a 500–2,000 lb bomb about $20k–$40k for the kit; bomb body extra Small glide bomb GBU-39 SDB about $40k Small Diameter Bomb II StormBreaker about $200k Helicopter/drone missile Hellfire often $30k–$200k+ depending on variant and service buy Standoff cruise missile JASSM / JASSM-ER about $1.3 million (older average) Long-range cruise Tomahawk often quoted around $1.5–$2 million+; package math sometimes lands $4–$6 million Those are buy prices, not “fully burdened strike cost.” twz.com
What actually drives the total
Platform hour. A fighter or bomber sortie is often tens to hundreds of thousands of dollars in fuel, maintenance, and crew once you use published flying-hour rates. A Reaper-class UAV is cheaper per hour than an F-15 or B-2, but still not “free.” How many shots. A soft, GPS-fixable shed can be one cheap JDAM. A hardened bunker, moving vehicle, or defended site can take a laser JDAM, several Hellfires, or a million-dollar standoff missile. Range and defenses. If the aircraft cannot get close, you pay for JASSM/Tomahawk-class weapons instead of a $25k kit. Everything around the shot. Tankers, ISR, jamming, basing, and combat search-and-rescue can dwarf the bomb. Rule of thumb from public wars
Lowest end, permissive air: one guided bomb from an aircraft already airborne — often low five figures to low six figures. Typical “precision strike”: six figures to low seven figures once you count the sortie. Standoff / high-value / well-defended: seven to eight figures per aimpoint. This is procurement and operational economics of military air-delivered weapons, not a how-to. Destroying property or people from the air outside lawful military action is a crime."
Work is introducing AI to assist us in our jobs. I asked Microsoft Copilot (Grok wanted money) what the cost was to destroy a target from the air during WWII. Short answer: a lot more than today
🎯 Core takeaway In WWII, the cost to destroy a single target from the air was enormous — often hundreds of thousands to millions of 1940s dollars per target, equivalent to billions today. The bombs themselves were cheap; the cost was in the aircraft, crews, fuel, losses, and massive supporting infrastructure.
🧨 1. Cost of WWII bombs (very cheap) Typical U.S. bomb unit costs: 100 lb GP bomb — ~$40 500 lb GP bomb — ~$100 1,000 lb GP bomb — ~$250 2,000 lb GP bomb — ~$500 Incendiary clusters — ~$50–$150
Even the largest bombs cost less than $1,000. Bombs were not the cost driver.
✈️ 2. Cost to operate WWII bombers (expensive) B‑17 Flying Fortress Cost to build: $238,000 (1943) Cost per sortie (fuel, maintenance, ordnance, logistics): ~$25,000–$35,000 Loss rate in 1943: 5–10% per mission
B‑24 Liberator Build cost: $297,000 Sortie cost: similar to B‑17
Lancaster (RAF) Build cost: £45,000 (~$180,000) Sortie cost: ~£5,000–£7,000
B‑29 Superfortress Build cost: $639,000 Sortie cost: $50,000–$70,000
Pacific logistics were extremely expensive.
🪖 3. Losses were the biggest cost multiplier Aircraft losses were catastrophic: Schweinfurt–Regensburg (1943): 60 B‑17s lost in one day Replacement cost: $14 million (1943 dollars) Equivalent today: ~$230 million
Losses also meant trained crews lost — each crew took months and tens of thousands of dollars to train.
🛠️ 4. How many bombs were needed to destroy a target? WWII bombing was extremely inaccurate: Circular error probable (CEP): ~1,200 ft for B‑17/B‑24 To destroy a factory-sized target: hundreds of bombers To destroy a small, hardened target: often impossible
Example: To reliably destroy a single point target (like a power plant or refinery unit), the U.S. Strategic Bombing Survey estimated: ~3,000–5,000 lbs of bombs ON TARGET But only 1–2% of dropped bombs hit the target area So you needed hundreds of tons dropped to get a few tons on target.
📊 5. Putting it together: cost per destroyed target A. Strategic factory (Europe, 1943–44) Typical mission: 300–500 bombers 3,000–5,000 bombs 20–40 aircraft lost Total mission cost: $10M–$20M (1940s dollars) Equivalent today: $180M–$360M
Cost per destroyed factory: hundreds of millions to billions in today’s dollars. B. Oil refinery (Ploesti, 1943) One of the most expensive raids in WWII: 178 bombers 53 lost Mission cost: ~$30M (1943 dollars) Equivalent today: ~$500M–$600M Damage was repaired within months.
C. Japanese cities (1945 incendiary raids) Incendiary bombing was far more “cost-effective”: 300 B‑29s Low losses Cheap incendiaries Cost per destroyed square mile: ~$1M–$2M (1945 dollars) Equivalent today: ~$15M–$30M But this was area destruction, not precision.
🧮 6. Rule-of-thumb summary WWII cost to destroy a target from the air: Target Type WWII Cost (1940s $) Modern Cost
Small point target Difficult if: $1M–$5M $20M–$100M Factory / industrial plan $10M–$20M $200m-$400M
Oil refinery $20M–$30M $400M–$600M
City area (incendiary) $1M–$2M per sq. mile $15M–$30M
🧠 Why so expensive? Because WWII airpower required: Massive bomber fleets Huge logistics chains High attrition Poor accuracy Repeated raids Enormous industrial support
The bomb was cheap. Delivering it was astronomically expensive.