Why were later Russian engines inferior to the NK-33?

Question

In terms of thrust-to-weight ratio the NK-33 engine developed for the Soviet N-1 rocket was by far the best in the world at the time, with a thrust-to-weight ratio of 137. But subsequent Russian rocket engines had significantly worse thrust-to-weight ratios (although better specific impulse); I'm thinking in particular of the RD-170 (TWR of 75) and its derivatives, the RD-180 (TWR 78) and RD-191 (TWR 89). Why were these later engines inferior, at least in terms of thrust-to-weight ratio (which is critical for first-stage engines)?

I can think of a few possibilities:

1. Later engines were over-engineered to make them more reliable? I don't know how to assess this. These engines never actually flew on the N-1, so the N-1's bad reputation should not have reflected on them. I've seen statements that the NK-33 engines were highly reliable because the designers learned from the failures of the previous N-1 engines. I also know that the Antares-100 launch failure was caused by an NK-33 and that Orbital ATK subsequently switched to another engine for the Antares-200; but as the engine in that case had been sitting in a shed for 40 years and refurbished by an American company, it's not clear whether its unreliability can be tied to the original design.

2. TWR was sacrificed for specific impulse in later engines? The NK-33 had a sea-level specific impulse of 297s; the RD engines mentioned above vary from 309s to 311s. My intuition from playing Kerbal Space Program is that this is an unfavorable tradeoff for first-stage engines, but I've not run any numbers for real-life rockets; and it's also possible that designers were reusing designs and/or components between first-stage and upper-stage engines.

3. Later engines had gimbaling capability, adding significant weight? When Aerojet refurbished the engines and added gimbaling, the resulting AJ26-62 had a 19% increase in engine mass. This suggests that gimbaling could explain a decrease in TWR from 137 to about 115, but not down to 75 or even 89.

4. Soviet design bureaus did not share technological improvements? I'm really hoping this isn't the correct answer since I'm hoping to learn about rocket engines rather than Soviet politics. And bolstering my hope, although the subsequent Russian engines were manufactured by a different design bureau, they had the same oxygen-rich closed cycle technology as the NK-33; if the rocket scientists shared that technology (which was believed impossible by Western rocket scientists at the time), why would they not have shared whatever other tricks were involved in getting such a light engine?

An additional note: The RD-* engines were significantly larger than the NK-33, so one might argue that TWR was sacrificed for per-engine thrust (perhaps because, following the N-1's failure, large numbers of smaller engines were not considered a viable strategy). However, I would (naively?) expect that economies of scale would tend to give a larger rocket engine better TWR, not worse.

Answer 1

TWR was sacrificed for specific impulse in later engines? The NK-33 had a sea-level specific impulse of 297s; the RD engines mentioned above vary from 309s to 311s. My intuition from playing Kerbal Space Program is that this is an unfavorable tradeoff for first-stage engines...

In the RD-170 case it's almost exactly an even performance tradeoff; propellant mass dominates the mass of the engines for most rocket stages, so even small differences in specific impulse can outweigh large differences in engine TWR.

5x NK-33 produce just 4% more thrust than a single RD-170. If we provide tankage for the rated 150 seconds of burn for the RD-170, and 144 seconds for the NK-33s, we get the same total impulse delivered over the course of the burn.

Assuming 6% of propellant mass for tankage, plus 200 tons of non-tankage-specific structure, upper stage, and payload, you get remarkably similar performance: the RD-170 produces 2840 m/s of delta-v, and the 5x NK-33 produces 2820 m/s (computed using the sea level specific impulse; the vacuum specific impulse gives the NK-33 the edge, although by a smaller margin).

Acceleration off the pad is almost identical: 1.25 g for the RD-170, 1.27 g for the 5x NK-33.

As you increase burn time, the propellant mass increases and so specific impulse becomes more important relative to TWR. 150 seconds is towards the short end of first-stage burns, and the -180 and -191 were rated for much longer burns than the -170.

Soviet design bureaus did not share technological improvements? ...if the rocket scientists shared [oxidizer-rich staged combustion] technology, why would they not have shared whatever other tricks were involved in getting such a light engine?

Sharing engineering techniques is one thing; getting another agency to adopt those techniques unchanged is quite another. Kuznetsov and Glushko were two different designers with two different philosophies; even with access to the same bag of tricks, it would be surprising if their engines weren't different in significant respects.

Later engines were over-engineered to make them more reliable?

This is a definite possibility. This wouldn't even have necessarily been a deliberate decision due to deficiencies in the NK-33 -- it could just be the result of one designer's gut instinct about how much safety margin to provide in the design.

Ultimately, though, you're setting yourself up for disappointment if you're looking for purely performance-based explanations for one engine being used instead of another. No rocket is ever built with performance-optimal parts. There are political considerations about which bureau/factory/company gets which contracts. There are cost and time considerations (which are often at odds) and development versus production considerations (which are often at odds). An engine that costs \$5M may be more desirable than an engine that performs 20% better at \$10M per unit.