The buffers are also counted as "absorbed" pollution. When a machine is destroyed, the pollution is released back into the environment and the statistics will show that the machine caused pollution.
Could this be the reason for the mismatching numbers?
Sorry I'm dumb and got confused.
Everything works as intended,Mk-3 filter cleans 300 pollution per min and consumes 15 reusable filters. So 20 pollution for 1 filter. https://imgur.com/a/Ihek1ZD