"From the mid-90's to the mid-2000's, significant advances were made in computer-based tools for closed-loop aero shape optimization, Multi-Disciplinary-Optimization (MDO) tools for doing complex multi-dimensional trade studies across scores of design variables, structural Finite Element Analysis (FEA) optimization for structural weight and stiffness. Much of this work was done as part of attempting to meet the requirements of the HSCT (High Speed Civil Transport) through its NASA-funded "HSR" element, and the follow-on industry internal R&D efforts toward designing other "concept planes" like the Sonic Cruiser, SSBJ concepts, and Blended-Wing-Body (BWB). The supersonic airliner concepts and SSBJ in particular, required being able to optimize not just the wing shape, but the wing, body, and nacelle shapes simultaneously, optimizing to find the best possible compromise between cruise aerodynamics, transonic acceleration, subsonic cruise, and low-noise takeoff and landing performance. The very thin, supersonic wings and tails, also provided significant weight reduction, aeroelastics, and flutter challenges that acted as catalysts for the improved structures tools. The MDO capability (still evolving) helped make the right kinds of trade-offs, i.e. "optimum compromises", in the design of the 787 and 777X. The improved CFD for air-loads prediction were coupled with FEA strength and stiffness optimization and these were used to guide direct optimization of aerodynamic shapes. The advent of practical direct CFD optimization of detailed aerodynamic shapes for 787 enabled the computer to find multiple incremental drag reductions that the "pressure matching" methods used on the original 777 could not. This optimization plus the development of very strong carbon fiber primary wing structures resulted in trade-offs that determined it was better to make the wing thinner while using less sweep to get a fairly high cruise Mach of 0.85--- the resulting thinner all-carbon wing was plenty strong enough in bending, but was more flexible than previous wings."

From the mid-90's to the mid-2000's, significant advances were made in computer-based tools for closed-loop aero shape optimization, Multi-Disciplinary-Optimization (MDO) tools for doing complex multi-dimensional trade studies across scores of design variables, structural Finite Element Analysis (FEA) optimization for structural weight and stiffness. Much of this work was done as part of attempting to meet the requirements of the HSCT (High Speed Civil Transport) through its NASA-funded "HSR" element, and the follow-on industry internal R&D efforts toward designing other "concept planes" like the Sonic Cruiser, SSBJ concepts, and Blended-Wing-Body (BWB). The supersonic airliner concepts and SSBJ in particular, required being able to optimize not just the wing shape, but the wing, body, and nacelle shapes simultaneously, optimizing to find the best possible compromise between cruise aerodynamics, transonic acceleration, subsonic cruise, and low-noise takeoff and landing performance. The very thin, supersonic wings and tails, also provided significant weight reduction, aeroelastics, and flutter challenges that acted as catalysts for the improved structures tools. The MDO capability (still evolving) helped make the right kinds of trade-offs, i.e. "optimum compromises", in the design of the 787 and 777X. The improved CFD for air-loads prediction were coupled with FEA strength and stiffness optimization and these were used to guide direct optimization of aerodynamic shapes. The advent of practical direct CFD optimization of detailed aerodynamic shapes for 787 enabled the computer to find multiple incremental drag reductions that the "pressure matching" methods used on the original 777 could not. This optimization plus the development of very strong carbon fiber primary wing structures resulted in trade-offs that determined it was better to make the wing thinner while using less sweep to get a fairly high cruise Mach of 0.85--- the resulting thinner all-carbon wing was plenty strong enough in bending, but was more flexible than previous wings.

"Very generally, yes, the computer-driven optimizations would seek ways to improve L/D not just drag or drag-rise. So they don't really use pressure distributions per-se as the figure of merit, rather the actual L/D that results from various shape changes. This means that in addition to trying to further reduce shock strength, they try to incrementally raise the net difference between upper and lower surface pressures thereby creating more "L" for a given "D". They also distribute the lift span-wise in a patter than does not follow the "ideal" elliptic lift distribution but rather try to create a down-wash field that allows the horizontal tail and fuselage lift to "trim" the airplane while minimizing the far down-stream wake. In a 2D airfoil sense, carrying lift farther aft on the wing (both to increase L and reduce shock strength) can cause thicker boundary layers, increasing the parasite "form drag", but the thicker boundary layers also have less local skin friction--- so the optimizer can seek the best L/D trade-off of the various drag sources (wave/shock drag+skin friction+span load induced drag+ tail and wake induced drag + wing pressure form drag), and of course do this in the presence of the nacelle and strut. It can also camber the nacelle and strut shapes, change inlet lip shapes to reduce inlet drag, and put local "blisters" or waves in wing surface, or change the shape of root fairings or wing tips to get a better combination of local pressure and the local surface slopes those pressures act on in the lift and drag directions (very much a 3D effect !). You say you have trouble picturing what would be advantageous for a full 3D optimization, but that is exactly the point--- the computer can systematically do the grunt work to investigate the effect of thousands of candidate shape variations, identify which have the most performance leverage, and recommend specific shaping trends that do not violate multi-disciplinary constraints (spar thickness for example)."

Very generally, yes, the computer-driven optimizations would seek ways to improve L/D not just drag or drag-rise. So they don't really use pressure distributions per-se as the figure of merit, rather the actual L/D that results from various shape changes. This means that in addition to trying to further reduce shock strength, they try to incrementally raise the net difference between upper and lower surface pressures thereby creating more "L" for a given "D". They also distribute the lift span-wise in a patter than does not follow the "ideal" elliptic lift distribution but rather try to create a down-wash field that allows the horizontal tail and fuselage lift to "trim" the airplane while minimizing the far down-stream wake. In a 2D airfoil sense, carrying lift farther aft on the wing (both to increase L and reduce shock strength) can cause thicker boundary layers, increasing the parasite "form drag", but the thicker boundary layers also have less local skin friction--- so the optimizer can seek the best L/D trade-off of the various drag sources (wave/shock drag+skin friction+span load induced drag+ tail and wake induced drag + wing pressure form drag), and of course do this in the presence of the nacelle and strut. It can also camber the nacelle and strut shapes, change inlet lip shapes to reduce inlet drag, and put local "blisters" or waves in wing surface, or change the shape of root fairings or wing tips to get a better combination of local pressure and the local surface slopes those pressures act on in the lift and drag directions (very much a 3D effect !). You say you have trouble picturing what would be advantageous for a full 3D optimization, but that is exactly the point--- the computer can systematically do the grunt work to investigate the effect of thousands of candidate shape variations, identify which have the most performance leverage, and recommend specific shaping trends that do not violate multi-disciplinary constraints (spar thickness for example).