I want to clarify/emphasize a few things in my answer.
#1: Ultimately, the core driver of development time is growing complexity.
More advanced threats means more advanced systems, which usually means more complexity.
- Your opponent gets a bigger radar, so you need a bigger one to see him first. You add jammers to shorten his effective radar detection range; he switches to IR sensors to augment his radar; you have to reduce your IR signature (buried/insulated engines, serrated nozzles).
- He coordinates with ground controllers and their huge radars; you jam his comms; he adds anti-jam functionality to his comms; you coordinate with AWACS to extend your own detection range and add standardized comms so everyone can talk to each other.
- You go low to hide your radar signature in ground clutter; he adds Doppler filtering.
- He gets IR missiles; you get IR jammers; he gets jam-resistant IR seekers; you get lasers to burn his seekers.
- You build a more maneuverable fighter; he gets super-maneuverable dogfighting missiles that can lock on at extreme angles.
These advanced capabilities aren't optional. They're necessary for survival. Even more capability is required to actually execute a mission.
- No missile approach and warning system? You might never see the missile coming.
- No radar jammers, no laser IR jammers, no decoys? Then his missiles are more likely to hit.
- No low probability of intercept functionality (radar and comms)? Then he could detect you from long range (rendering your stealth less useful) or jam your radar.
- No ground-mapping modes? Then you have to spend more time in hostile air space looking for a target. Or you have to rely on someone else to find targets for you (now you need two aircraft).
- No integrated ground targeting pod? Then you need an external pod, which is draggier and increases you RCS.
- No stealth? No jamming? Have fun starring at long range radars and SAMs.
- No jam-resistant comms? Have fun talking to yourself.
#2: So why not design everything incrementally?
Why not fly a quick n' dirty design, find the problems, and then fix it in the next block?
First, we still do that (to a degree). The F-22, F-35, and B-21 all have upgrade paths.
Second, that can be more expensive and time consuming that designing everything at once.
- A \$1 flaw in design costs a \$10 in production and \$100 in the field. When aircraft are already so expensive, the flaws will also be expensive to fix.
- When aircraft are so complex, purely serial design/development would take forever.
- Low-rate production is inefficient/costly. Stopping production between blocks is even more costly.
But designing everything simultaneously is more complex (more balls to juggle).
So there's a balance between serial and concurrent development.
#3: Development time is counted differently
Better testing and greater safety imposing more time demands. - [from the updated question]
It's not only that modern testing is more stringent/complete/safe, but that modern development time is counted differently.
- F-16 development time is longer than it appears because the F-16 essentially entered service before testing was complete. (Remember how the tail had to be enlarged by 25% to address deep stall? That fix didn't happen until Block 15, after 329 airframes were already built.) By modern standards, the early F-16s would still be in testing.
- And USAF dissatisfaction with the F-16's (and F-15's) engine led to a bitter dispute with PW, leading to the alternate engine program. Eventually, GE engines power most F-16s today.
- The F-14 is also imfamous for its bomber-derived engines, which performed poorly at high AOA.
So it's not just about safety, but also basic functionality.
In comparison, there's a couple hundred F-35s around today (~300, ~210 in the field, ~90 coming out of production), more than the F-15C/D (~180), F-22 (~180), or F-15E (220+) fleets. (Some are already in service with the US Air Force and US Marine Corps, others are wrapping up testing, and still others are stateside training the initial cadre of pilots for various partner nations.) While most of them aren't in service, they're already at a more "advanced" state of testing than the F-16 was after it had already entered service.
Of course, besides safety and thoroughness, more testing/development is necessary because more complex aircraft have more things to test/develop.
4: Multiroles can save money and can provide more capability
I still have great difficulty in accepting that... Having fewer airframe types saves cost and/or development time, this seems evidently incorrect. - [from the updated question]
First, if two aircraft are similar enough, you might use the same aircraft to serve both roles. You develop one aircraft instead of two, theoretically cutting your development costs in half. Or you can use those savings to design a better plane. In practice, you might need to make some mods to suit different roles/customers, but the airframe and systems are [hopefully] similar enough to yield net savings (or a better aircraft).
This isn't unprecedented. The YF-16 and YF-17 both fought for the same US Air Force program, but the USAF took the F-16 and the Navy developed the YF-17 into the F-18.
Second, the F-16 already does this. It already performs a variety of roles previously flown by several different types of aircraft. It's used for everything from close air support, interdiction/deep air support, and suppression of enemy air defenses, to air superiority and recon.
The Super Hornet is also a potent multirole platform. Even the F-15E Strike Eagle, built for long range interdiction, has basically the same air superiority capability as the F-15C. Even the F-14 earned its "Bombcat" moniker. In fact, most modern and [surviving] legacy fighters are multirole: F-15, F-16, F-18, F-22, F-35, Su-27 family, Gripen NG, Rafale, Typhoon, J-10, J-20, etc.
Multiroles are flexible and increase total available capability. Example:
- Say Plane A (air superiority) does "100" air superiority and 0 ground attack.
- Say Plane G (ground attack) does 0 air superiority and 100 ground attack.
- Say Plane M (multirole) does 75 air superiority and 75 ground attack.
- Say you buy 100 planes. You have two options:
| Option 1 | Option 2
| single-role | multirole
number of Plane A (100/0) | 50 | 0
number of Plane G (0/100) | 50 | 0
number of Plane M (75/75) | 0 | 100
air sup. capability | 5000 | 7500
ground. capability | 5000 | 7500
The multirole Plane M may not be as good as A or G in any specific role, but you get more capability overall. But if you build multiroles, you potentially have 100 planes available for each role.
This works because missions change. On Day 1 you might need a lot of air superiority, and on Day 2 you might need a lot of ground-attack. Well, with only single-role fighters (option 1), half your fleet is useless on both days. But with multiroles (option 2), all of them are useful on both days. (Missions can even change on the same sortie, eg an interdiction group defends itself against opposing fighters before proceeding to their target. F-16 strike fighters that can defend themselves can be more efficient than always assigning "just in case" F-15 escorts.)
This is even more true if you have more roles. If you have 10 different roles, you can only build 10 planes (average) for each role (in the single-role option). So up to 90% of your fleet might be useless on a particular day. (This is obviously an extreme example.)
But even at the least-optimal point (max A, and max G), the multiroles still provide good capability (7500) compared to the single-roles (10,000). The single-roles' capability will vary between 5,000 and 10,000 (depending on day), whereas the multiroles can always bring 7,500.
There's an exception. If you know beforehand you always need "5000" worth of air superiority, then yes, it's more efficient to build 50 air superiority fighters. But A) it's hard to know that beforehand (predicting the future), and B) that number will change over time regardless. But if you know you'll always need at least "1000" air superiority, then it's more efficient to buy 10 air superiority fighters (actually more than 10, depending on expected use). In practice, you'd build at least 10 air superiority fighters and give then some multirole functionality.
(There's another exception. This model assumes all Air Superiority or Air to Ground capability is identical, regardless of source/type. Obviously this isn't always true.)
Notice the break-even point in the example above. Plane M was 75/75, but if it was less capable (50/50), then Option 1 and 2 both offer the same available capabilities. So the multirole plane must be competent in its typical roles, otherwise the fleet composition is inefficient. On the other hand, the multirole plane doesn't have to excel at its typical roles, it just needs to be competent.
Thankfully, multirole fighters can perform a variety of roles very well just by swapping payloads. Take the F-16 for example. Need defensive counter air? Load up the AMRAAMs. Need CAS? Grab nav/targeting pods and load some JDAMs/LGBs. Recce? Photo-recon pod. SEAD? HARMs, decoys, and jammers. Anti-armor? Mavericks and SDBs. Antiship? Harpoons. Don't forget the fuel.
Recall how missions can change mid-flight. Those self-escorting F-16s are performing both roles simultaneously, say 25/75, so their total capability being used is 100, and fleet capability is 10,000 (equal to the single-roles on their best day or 2x the single-roles on their worst day).
The F-35 can perform multiple roles simultaneously. They can escort themselves, find targets themselves, strike a ground target, jam enemy aircraft and ground radars, and even take out enemy SAMS, all the while acting like mini-AWACS/JSTARS for friendlies. Say that's 75/75/75/75/75/75, or 450 total per aircraft and 45,000 for the fleet.
That's part of the value of multirole stealth fighters. They need far less support.
(Image source: Beyond the "Bomber": The New Long-Range Sensor-Shooter Aircraft and United States National Security - Lieutenant General David A. Deptula, USAF (Ret.), 2015.)
That's partly why you see the F-35 winning 20 vs. 8 fights and racking up >20:1 kill ratios at Red Flag:
#5: [Multirole example] The F-35
[This section really deserves its own question/answer, but I'll abbreviate it here.]
The F-35A (\$94.6 million (FY16\$)) costs less than the Typhoon, Rafale, Gripen NG, and Super Hornet. In 2019, it'll cost a touch more (\$80 million) than a Block 50 F-16 (\$65-\$80 million).
Compared to the F-16, Hornet, Super Hornet, Harrier, A-10, Rafale, Gripen, and F-117, the F-35 is more capable in almost every relevant aspect. It has the best radar, best EW suite (probably), and best IR sensor suite. It has the best range and carries the most payload. It has the "superb low speed handling characteristics and post-stall manoeuvrability" of the Hornet and the acceleration of an F-16 (a 'Hornet with four engines,' remarked one pilot). At equivalent combat loadings, the F-35 is faster and accelerates/climbs more quickly. Sensor fusion and network integration provide excellent situational awareness (perhaps second to only AWACS) and greatly facilitate coordinated strikes.
Even versus the Raptor in air-to-air, it's not clear the F-22 would dominate. Some F-35 systems (the APG-81 radar, ASQ-239 EW suite, AAQ-37 DAS sensors) are direct upgrades of F-22 systems (APG-77, ALR-94, AAR-56). The F-35 radar ground mapping and targeting capability was actually backfitted onto the F-22. The F-35's EW suite has detected and jammed the F-22's radars. It's also twice as reliable and four times cheaper than its predecessor on the F-22. And the F-35's EODAS evolved the F-22's AAR-56 into much more than a missile launch detector, also providing ground-fire geolocation, weapon cueing, and situational awareness IRST (the famous "see through the cockpit floor"). Test pilots and USAF officials have also remarked that the F-35 is stealthier than the F-22.
The F-35 also has an integrated IR search and track system and helmet mounted cueing system. Both were canceled on the F-22. The F-35 will also carry 6 internal AMRAAMs (same as the Raptor) or 4 internal MBDA Meteors.
- F-16: 2 years for demo (1972-74), 4 years to first flight (1972-76), 8 years to service entry (1972-80)
- F-35: 5 years for demo (1996-2001), 10 years to first flight (1996-2006), 19 years to service entry (1996-2015)
Overall, F-35 development ostensibly took 2.5 times longer than F-16 development. Given the tremendous amount of new technology, it's notable that development time was comparable to other modern fighters, which also average 20 years (see far below) despite being far less technically challenging.
By running a joint program, the JSF eliminated a lot of potentially redundant system RDT&E and reinvested those savings into more advanced systems (and more systems). Eg, instead of designing three different radars, they designed a single (more advanced) radar---avoiding reinventing the same radar twice over, avoiding three separate test and validation programs, and avoiding managing three separate programs.
#6: Minor points/additions
I still have great difficulty in accepting that...
A reason would be the complexity of the problem, which is very often quoted. All aerospace problems are extremely complex, that is nothing new. But now nobody can tackle them anymore? That was the basis of my question.
I think the complexity has simply grown faster than the tools' ability to keep pace.
I think with modern tools and processes, it would be very easy to out-design any plane from before 1980.
You only do these new programs every so often, there's so much of a
backlog of stuff you want to get done---you throw too much at it. You
want to do all the new technology, all the new manufacturing
processes, new tools, new everything all at once. It makes the
complexity exponential. [So] You need to find things to advance
outside of these programs to get yourself ready for these programs, so
when the opportunity comes up... the technologies [will already be] in
the market, so you know what you're doing so you can execute...
instead of "it's been so long that we need to start at ground zero
[with] a new program plan, a new organizational structure, a new
teaming"---all that at once makes it hard to execute. - David Kusnierkiewicz, Mission Systems Engineer (NASA), Johns Hopkins University Applied Physics Laboratory. Speaking at AIAA ("Whatever Happened to the Four Year Airplane", @31:00)
Modern aircraft are vastly more complex
- An F-16 weighs as much as an B-25 (9 tonnes). An F-22 (20 tonnes) weighs 60% as much as a B-29 (34 tonnes).
- The F-16C had 150,000 lines of code. The F-22 had 2 million lines of code. The F-35 has 8+ million. The F-35's logistics system has 24 million lines of code.
- Automation/software is basically narrow AI, replacing several extra crew members: radar terrain mapping, analyzing ground targets through multiple sensors (optical, IR, radar), automatic target recognition, missile approach and warning, electronic attack, defensive countermeasures, etc.
- Even engine start up is simple, "requiring just three switch selections, one each for the battery, integrated power pack and the engine... Initiated by a button in the cockpit, the [vehicle systems built in test (VS BIT)] self-tests almost every function imaginable on the aircraft... After 90 seconds, if there are no problems, the aircraft declares itself ready for flight" (Mark Ayton).
- In comparison, the SR-71 is practically just an airframe with a camera and jammer.
- Systems are much more complicated than the basic structure:
- A "stealthy" low probability of intercept AESA radar, distributed aperture system for missile approach and warning and 4pi steradian observation, helmet mounted display, electro-optical targeting system for IR search and track and ground targeting, electronic warfare suite (which, in the F-35, is reputedly much more complex than even the radar), "stealthy" low probability of intercept and high data rate comms, electro-hydrostatic actuators (each actuator is self-contained rather than relying on an aircraft-wide hydraulics loop), and of course stealth (from the LPI comms, LPI radar, skin, to the nozzle).
- Sensor fusion. Then you must fuse all those sensors and comms together and integrate with the rest of the fleet so that everyone sees the same picture. If a lone drone spots a distant fighter/tank/missile, everyone immediately sees it as well. Sensor fusion is responsible for much of the system/software complexity. But you cannot divorce the systems from the aircraft. They are part and parcel of the 5th generation experience... and thus responsible for the incredible capabilities necessary for the future.
- In other words, much of the complexity is necessary to deliver the requisite capabilities.
More development and testing before service entry
- Modern safety standards
- Earlier aircraft entered service before many of their iconic systems were finished. In modern practice, the first few hundred F-16s would still be in development/testing, rather than pushed into service before even the airframe and control laws were completed.
- For example, in just its first few years of service, the F-16 had 50 crashes. It was nicknamed Lawn Dart for a reason. In contrast, the F-35 has had zero crashes and only two Class A mishaps, once when they flew too hard without first breaking in the engine and once when they tried starting the engine with excessive tail wind.
The physics are better understood
- The 50s, 60s, and 70s saw a flurry of new designs as new frontiers were explored. These were poorly understood, so as basic knowledge accrued, older designs rapidly became obsolete. And since airframes were relatively simple and manpower high, designs evolved incrementally but continuously (rather than in discontinuous leaps today). Opposite NATO, the Soviets were doing the same and thus engendered continuing competition.
Twenty years is pretty typical for modern aircraft development. The F-22, F-35, Typhoon, Rafale, Hornet (including 8-10 years of YF-17 development), and even PAK FA all took (or will take) about 20 years from start to IOC.
A quick note on the SR-71. It wasn't developed in two years. It was derived from the Lockheed A-12, which flew before that SR-71 mockup was even shown. The A-12 program began in the late 50s, and the J58 engines started development even before that, initially for the P6M strategic flying boat, which first flew in 1955.
A quick note on O2 issues. A lot of high performance jets have had recent hypoxia issues, not just the F-35, including the Hornets, Super Hornets, T-45s, and [infamously] the F-22s. The F-15, U-2, and SR-71 also had their fair share of O2 issues.
Just for fun: AIAA AVIATION 2015, "Whatever Happened to the Four Year Airplane"