Many hard-core automotive enthusiasts (and not a few curmudgeons) decry the dwindling availability of conventional manual transmissions in today's cars. There are a number of reasons for this, many of which relate to human factors ("are there enough buyers of a manual, in this car, to make such an option profitable") , but there is another factor that you may not have considered: engine exhaust emissions. And how, you ask, could the type of transmission have an effect on emissions? The one-word answer: transients. Yes, the very act of shifting your beloved old-school manual transmission makes life harder on OEMs than developing something as complex as an automatic transmission.
But before we dig into the effect that transients have on emissions, let's discuss a little background. Having developed the application code for, and calibrated, automatic transmission controllers, I can relate to you that this is no small affair, but it pales in comparison to engine controller development. An engine controller typically has about three times the code (upwards of 8 megabytes of executable application code and calibration data which, for an embedded controller, is a lot), with more algorithmic control functions and much more in the way of "look-up" based calibration. Transmission control software, particularly the projects I have been involved with, are more "state machine"-type of applications, with algorithms for things like fluid temperature compensation, clutch-pack engagement control and similar gradient-based functions. Engine controllers, on the other hand, have many more gradient functions, such as spark timing, fuel-air management, fuel vapor management, speed control and a host of others.
So...the scale of working with engine software is very large, even for the simplest, most steady-state control items. This complexity has been partially mitigated through a quiet, very nerdy revolution: the advent of model-based ECU development. Model-based development allows an engineer, working within their core competence, such as mechanical engineering or chemistry, contribute to electronic control system development by modelling the desired behavior, and then using a tool to turn the model into microcontroller code. This process has a number of benefits, including the inclusion of non-programmers into the development team and, generally, better code.
This background information is cool (and by "cool", I mean industry-insidery-boring), but how does this relate to transients, and why does it mean that I can't get a manual transmission in any car that I darned well please? The answer is that transients are difficult to model, difficult to cope with in code and create a surprising amount of tailpipe emissions. And a human, shifting a manual transmission, creates rather unpredictable transients, which create more emissions than a nice, predictable automatic transmission.
Today's automatic transmission control systems, whether for planetary or for "dual-clutch" equipment, measure inputs such as driver torque demand, RPM, temperature and vehicle speed which allow the engine controller to derate the engine, if needed, and control the fuel-air ratio (F/A) in such a way as to prevent a spike in emissions. Even "floppy-paddle"-equipped automatics, with their ability to accept input at the driver's whim, still have a little processing time (partially because of now-ubiquitous "drive-by-wire" throttle control systems, which isolate engine air intake from the driver's foot) time to cope with the upcoming change in engine load, and the resulting changes need in fuel and spark. Not so with a conventional manual, where a shift can come at an time and without the same degree of "foreknowledge" of the rate of load change.
This fact makes conventional manual-equipped cars more labor-intensive to develop, because you can only model and dyno-test so well and, at the end of the day, you're going to have to have people drive around with instrumented cars, tweaking the calibrations and retesting the results. The obvious question that most people ask during this discussion is "but it's still easier than than calibrating an automatic", to which I respond "it's additional work, because it's not an either/or situation...the OEM is going to offer an automatic AND a manual, not the other way around".
The other, and larger problem, is that manual-equipped cars, based on the above, will produce more tailpipe emissions, all other things being equal. This creates additional difficulties for OEMs due to the fact that cars must not only pass "per-car" standards, but also contribute to a "fleet average", which is exactly what it sounds like - an average "pollution component per mile" calculation, where each model and drivetrain combination influences the average. Combinations that are at the higher end of the per-vehicle limits drive the average upwards, and typically need to be offset somewhere else...which compromise the combinations that get earmarked to help offset the "dirtier" cars.
I think OEMs that make the effort, and bear the expense of development of manuals are to be commended, and I would suggest that people that love them should buy them while they can - it's probably just a matter of time that the return on investment will demand the extinction of row-it-yourself cars.