I can't get even similar irradiance values from a lamp.

Dear Radiance users,

I'm writing here for third time and I always sorry for just asking.
I'm modeling a small box has a several bared cylindrical fluorescent lamps
(actually UV lamps) inside and want to know the irradiance level on the
surfaces of the box. The cylindrical lamps has 32mm in diameter and 540mm in
length and The radiance of 44W/sr/m2 calculated with the UV output of the
lamps (7.5W) was applied to the lamps. Before calculating the small box
model I checked irradiance values with an experiment measuring UV intensity
at the distance of 1 m and compared that with the same simple RADIANCE
calculation and the result was reasonable. However, the calculation
results are higher than those of experiments by 2~4 times with the small box
model. The small box is not simple but not so complicated. I have modeled
the lamp both as a cylinder and 72 polygons but there were not so much
changes. Only one channel was used to calculate irradiances of UV from
modeling to issuing results and only diffused reflections were assumed.
Geometric errors or abnormal radiance distributions of lamps could be
assumed to cause the discrepancy, but 2~5 times are too large
Here is my rtrace options.

cat pts_sensor.pts | rtrace -I -oov -ar 128 -ad 512 -as 256 -ab 2 -ds 0.02
uv.oct > uv.dat

It would be appreciated if anyone give me advice.