I successfully calibrated a BME680 last night using the BSEC 2.x library and bench-tested the sensor. I was able to calibrate by having my office window open for several minutes, and then closing the window. Shortly after the window was closed the sensor was in a fully calibrated state.
Two concerns have arisen from this bench-test.
1. I am a little concerned about the proprietary nature of this library since the algorithms for BSEC_OUTPUT_BREATH_VOC_EQUIVALENT and BSEC_OUTPUT_CO2_EQUIVALENT do not appear to available anywhere public. Nevertheless, for rapid prototyping of this project it will still remain valid.
2. The calibration time is a bit long and somewhat nondeterministic. While the sensor continuously calibrates, it does so using data from the past 4 days (in the current configuration). This raises questions about the accuracy of the sensor in an dynamic location/environment rather than one which is in a static location. A custom hysteresis algorithm may be required to compensate for these environmental shifts.