United States, as well as many of its allies, have always looked towards increasing range of combat as much as possible. Unfortunately BVR combat is, conceptually, operationally and technologically, a massively complex affair. BVR theory states that future air combat will be comprised of large "missile truck" aircraft flying at supersonic speeds, launching radar-guided missiles at targets that are way too far to be identified visually. This has resulted in development of aircraft that are very heavy (most weigh about 16.5 tons (15 metric tons) empty – for example, Tornado ADV weights 16 tons (14.5 metric tons), and F-22 weighs 21.7 tons (19.7 metric tons), carry large numbers of missiles, and are far more expensive and much less reliable than aircraft with a bias towards visual-range combat.
MISSILES
In order to pull as tight a turn as a fighter aircraft, a missile has to pull a total g-force that is the amount of g's the aircraft can pull multiplied by a factor of the difference in their speeds squared. Thus if a missile travels at Mach 3 and the fighter aircraft travels at Mach 0.6 (corner speed of many modern fighters) and can pull 9 g maneuvers, then the missile needs to pull 225 g to match the fighter's turn radius, or 100 g if the fighter is travelling at Mach 0.9. Now if the missile is fired outside of the ideal position, it has to maneuver in order to point its nose towards the target, thus lowering the probability of a kill; there is also a danger of targeted aircraft simply flying out of missile's field of view. This danger is still present with active-seeker BVR missiles. In BVR, an AIM-120 travels at Mach 4, and can pull 30 g within its NEZ, yet it would need 400 Gs to reliably hit a modern fighter which is maneuvering at a corner speed of Mach 0.6, or 178 Gs if target is still at standard cruise speed of Mach 0.9.
Main problem with evading missiles is their speed, which makes timing somewhat difficult. However, with a missile closing at 1200-1400 meters per second (in the best case), at 20 kilometers, this means 14-20 seconds to reach the target for a BVR missile, or 20-23 seconds for IR missile.
Now consider the missiles actual flight path. When a rocket powered air-to-air missile is fired at a target it delivers the same amount of thrust over a certain period regardless of the tactical scenario. If the target can be reached without the rocket motor burning out, or shortly after it does so, the missile will have a high-energy state during its terminal attack phase. This will allow it to maneuver very hard. If the target is farther away, the missile will usually climb to a high altitude while its rocket motor is burning and then coast on its built-up energy with gravity on its side until it reaches the terminal phase of its flight. If the target isn't too far away, and the missile is still above it, it will dive down on the target in an attempt to maximize its ability to make hard maneuvers. The longer the shot, the less energy the missile will have for its critical terminal phase of flight.
Consider the U.S. AIM-120 AMRAAM. It is regarded as a fire-and-forget weapon, and it does have a mode to do just that. Basically it takes the targeting data from the aircraft's radar and calculates where the target "should be" when it arrives in the target area.
It then flies out to that area using its own inertial navigation system. Once there, the missile's small radar seeker, which has far less range and scanning capability than the radar on the fighter that fired it, starts to look for the bad guy. If said bad guy is within the AMRAAM radar's cone of detection it can lock on and attack.
The problem is that at intermediate and long ranges the fire-and-forget performance is abysmal. If the target is not where the missile thought it would be, within a limited cone of the sky, it's a miss. As such, this mode is really for shots taken at close ranges where there is less flight time in which the enemy can change course, altitude and tactics.
The only way to counter this is by the fighter aircraft that launched it sending it mid-course updates as it flies out to the target. As it goes along its way, and as the range between the missile and the target decreases, its ability to predict where the target will be improves as it has much more recent telemetry to rely on. Ideally the fighter will provide updates to the missile until it locks its own radar on the enemy target.
While this works it means that the pilot has to keep his radar pointing towards the bad guy and continue sending radar data to the missile to improve its chances of a kill, but that exposes him to the enemy as range between him and the target decreases.
AIRCRAFT
Speed was life in air combat until the jet age. The pilots who became Air Force generals in the Fifties had learned their trade in the Thirties when speed was the most-desired quality in a fighter. Thus when the requirements for new fighter aircraft were written in the Fifties, those generals made sure that a higher top speed was part of the specification. The Generals got their supersonic aircraft which then flew in Vietnam. When the Air Force in the late Sixties accumulated the flight data from several years of Vietnam War air combat, they found that all aircraft had accumulated just minutes at Mach 1.4 and only seconds at Mach 1.6 out of more than 100,000 combat sorties. Never was even Mach 1.8 flown in aircraft which had been optimized for Mach 2.4 (F-104, F-105, F-106A, F-4D/E and F-111). The WHY is simple physics: the shape of the turn rate vs Mach number relationship for an aircraft. In combat, each pilot has the tendency to fly his aircraft so as to maximize his turn rate. He thus gains angular position on the enemy which, in turn, may permit a missile launch or a gun firing. It can be seen that the pilot's urge to maximize his turn rate will unfailingly drive his Mach number to about 0.7. Thus, if the pilot is going to join in combat, his speed will inevitably drop to subsonic speeds. Even if the turn rate is held constant while increasing the speed, the turn radius and load factor increase, bringing with it increasing problems of keeping the enemy in sight.
The second reason given in the study is the dramatically smaller combat radius once the aircraft starts to fly at supersonic speeds. Even for flying into the combat arena supersonic speed was rarely advantageous. Northrop studied a multitude of intercept cases and found that speeds above Mach 1.1 were almost never helpful because they curtailed the combat radius severely.
Flying at Mach 2+ requires heavy and complex air intakes, a heat-resistant structure, high wing sweep and heavy, low-bypass engines. This all degrades the combat qualities at high subsonic speed, which was where those aircraft were used the most. Building into them the capacity for Mach 2+ made them worse for what they were actually used for.
COMBINATION
There are also other problems such as reliable IFF at long distance, dangers of using active sensors in combat, increased weight degrading performance, cost factors and complexity penalties making it very difficult to maintain and repair BVR systems in the field, as well as training penalties caused the by aforementioned penalties on weapons system.
IFF problems, the only reliable IFF method is visual one, especially since pilots often turn IFF transponders off to avoid being tracked. Visual IFF, unless assisted by optical sensors (be it camera or IRST), usually requires two aircraft to approach within one mile or less (sometimes as close as 400 meters), whereas minimum range of AIM-120D is 900 meters. But even when assisted by visual sensors, it may not always be reliable, as opponent may be using fighters of same type or at least of very similar visual signature.
Active Sensors, are outright suicidal in combat. Aircraft using active sensors will be quickly detected and targeted by modern defense and EW suites, and a unique radar footprint may even allow for BVR IFF identification. This can allow passive aircraft to launch BVR infrared or anti-radiation missile, and/or to use data acquired to achieve optimal starting position and speed for following dogfight. Only countermeasure is to turn radar off and rely solely on passive sensors. IRST is especially useful here, as while air temperature at 11 000 meters is -56 degrees Celsius, airframe temperature due to air friction can reach 54,4 degrees Celsius at Mach 1,6 and 116,8 degrees Celsius at Mach 2. It is also very difficult to impossible to jam, and offers greater angular resolution than radar. Result is that flying from cloud to cloud is still a viable combat tactic; but it is not perfect either, as clouds are not always present and may not be close enough for aircraft to avoid detection.
Weight Difference. The Gripen C weights 6622 kg empty, compared to 19,700 kg empty for F-22; F-16A weighs in at 7076 kg compared to 12,700 kg for F-15C. It can be seen that WVR fighters are significantly smaller and lighter than contemporary BVR fighters. And with cost of $6645 USD per kg, Gripen C is significantly cheaper per unit of weight than F-22 which costs $13,300 USD per kg, whereas F-16A costs $4240 USD per kg, which when compared to F-15C's $9921 USD per kg gives similar ratio to F-22/Gripen one. Bigger radar – focus of the logic – required bigger airframe, which in turn required bigger engines. Both weight and complexity spiraled upwards, creating fighters that were costly, flew very few sorties and had maneuvering capabilities more typical of strategic bombers than of fighter aircraft – logic being that they will not have to maneuver, as they will destroy the enemy far before it comes to the merge. So we have the F-22 and F-35, both of which are utterly expensive and maintenance intensive, and latter of which is in its major characteristics more similar to century series than modern fighter aircraft. F-35 in itself is utterly incapable of handling itself in close combat due to large weight, high drag, high wing loading and low thrust to weight ratio. It can also carry at most 4 BVR missiles in internal bays. With this in mind, claims by manufacturer that F-35 is 4 times as effective in air-to-air combat as next best fighter in the air would require probability of kill for BVR missiles of 80-90%, and opponent's complete inability to engage F-35 itself at BVR range. Track record of BVR missiles to date as well as development of infrared BVR missiles and long range QWIP IRST sensors mean that any such assumptions are nothing more than wishful thinking on part of sales department and high technology addicts.
Training, BVR-oriented aircraft, low in number, hugely complex, extremely expensive to maintain and operate simply cannot be used for training often enough. While technologists typically counter this argument by pointing to increased ability of simulators, that argument is not realistic: simulation is never perfect, as quality of the end result is never better – and is often lot worse – than quality of data used to compute it. Simulators often misinterpret reality, and support tactics that would get pilots killed in real combat. Further, simulators cannot prepare a pilot for the handling of shifting g forces encountered during both dogfight and BVR combat maneuvering.