It would be helpful to know if you are running Hauptwerk IV or V, since the advanced audio routing changed a lot between versions. Also, there is a difference between the Advance and Lite version of V. It would be a good idea for you to download the Hauptwerk V user guide, since it goes through the audio setup in detail (
https://www.hauptwerk.com/documentation/). On page 184, the topic is "Audio routing and impulse response reverb part 2 (Advanced Edition only): multi-channel audio, multi-channel virtual acoustics, and routing concepts for advanced use" (though the chapter recommends reading the previous chapter first).
It seems as though currently you are using a 2 channel set, where the left channel is going to every left speaker, and the R channel is going to every R speaker. That means that you have 8 identical Left channels speakers, and 8 identical right channel speakers. Hauptwerk sees this as 16 speakers, 8 of which are getting an identical signal from channel 1, and 8 of which are getting an identical signal from channel 2.
For the advanced audio routing in Hauptwerk V advanced, for a 2 channel set, you would route all of the ranks to one of the primary busses, which would then divide the notes among different busses, then send it to the mixer where you can assign the same channel to multiple different speakers. By default, all audio goes to Primary bus 1, which is then sent to the Master Mixer bus 1. However, in the mixer screen, you can see that there is the "perspective mix: output perspective... options for the Master mixer bus 1. Each one of those options (Front1/main, Front2/upper, Rear1/main, and Rear2/upper) can be assigned to a separate speaker pair. Now, on the "Rank Routing to Audio Mixer Bus Groups" screen, you can assign each rank to feed the four perspective presets in a "Static: Cyclic within octaves, Octaves cycled, ranks cycled" manner, which means, for instance, when you play a 3 note chord, a different speaker will play each note, with the L stereo samples being played on the left 4 speakers, and the R stereo samples divided among the R 4 speakers.
This should significantly cut down on intermodulation distortion (making it sound less muddy), while preserving the quality of the 2 channel stereo image. Basically, instead of sending all three notes to all 8 speakers in stereo at the same time, it sends one left note to 3 different left speakers, and one right note to 3 different right speakers at the same time.
It is quite confusing when you first start out, but it gets easier with time. The manual takes a long time to get through, but it makes sense when you get the hang of it. Another forum user posted a nice tutorial on a separate topic, but I am having trouble finding it again. Let me know if you have any further questions.