IS there anyway we can simulate the raw data for bme680 sensor from somewhere? or we need to use the sensor itself? Can anyone help?

The BME680 is a gas sensor, the sensor response depends on the environment. If you like to simulate the sensor you need to build a mathematical model, take into account the environment where the sensor is located, and therefor you need data.

You need to use the sensor itself to generate this data.

If you have some samples of that (the more the better), you can try to see if any known statistical distribution can fit the data (Poisson, Gaussian, Gamma …). If so, you can fit the distro and get some of the required momentum to identify the distro (such as mean and std), once you have done that you can generate pretty reliable data.

I doubt if some samples is sufficient. It all depends the target accuracy of the mathematical model. I have no experience with this type of sensor, but low cost air quality sensors can have, for example, some non-linearity and cross sensitivity.

Buy a couple of sensors, start collecting data, get good understand of the data and problem you target and improve the model.