There are basically 3-4 main ways of determining fluence rate and thus fluence (UV dose) - biodosimetry, spectroradiometry, actinometry, and computer simulation.

Biodosimetry is the gold standard for UV reactor validation.

See Protocol for the Determination of Fluence (UV Dose) Using A Low-Pressure or Low-Pressure High-Output UV Lamp in Bench-Scale Collimated Beam Ultraviolet Experiments from IUVA: https://iuva.org/Guidance-Documents

Overview: https://iuva.org/resources/Resource%20Documents/Whitby-Uniform%20Protocol%20for%20Wastewater%20UV%20Validation%20Applications.pdf

It doesn't suffer from the fundamental issues present in the other methods, as the biodosimetry microbes follow the same path as those in the real wastewater you want to test. I think this is the best method for your situation.

Spectroradiometry probably wouldn't work in your case as you would need the place the sensor in many locations to map out the reactor, not to mention the absorbance needs to be accounted for.

Actinometry might work for your sitaution, but the water needs to be moving so that the actinometry solution near the lamp is not quickly saturated. Since the absorbance doesn't happen away from the lamp, it's difficult to convert the fluence near the lamp into the fluence throughout the entire reactor. Actinometry works best in a petri dish setup, where factors can be calculated to correct for these issues. See:

https://doi.org/10.1061/(ASCE)0733-9372(2003)129:3(209)

Article Determination of the quantum yield of the potassium ferrioxa...

Article Practical Chemical Actinometry – A Review

Computer simulation could work of course. See: https://www.comsol.jp/paper/virtual-prototyping-of-uv-led-based-point-of-use-water-disinfection-unit-66331