Huge difference between Positive Temperture and Negative Temperature

I define a positive temperture nonlinear profile and a negative temperture nonlinear profile on the same beam as follows, the absolute values of positive and negative temperature profile are the same:

LC NO TYPE TITL
9 TG Positive
LET#T_C +12.5 +3.35 0
LET#Z_C 0 0.1[m] 0.4[m]
LOOP#i T_C
BEAM FROM TO TYPE EZA PA REFT
‘GRP’ 21 T #Z_C(#i) #T_C(#i) -Z
‘GRP’ 22 T #Z_C(#i) #T_C(#i) -Z
‘GRP’ 23 T #Z_C(#i) #T_C(#i) -Z
ENDLOOP
LC NO TYPE TITL
9 TG Negative
LET#T_C -12.5 -3.35 0
LET#Z_C 0 0.1[m] 0.4[m]
LOOP#i T_C
BEAM FROM TO TYPE EZA PA REFT
‘GRP’ 21 T #Z_C(#i) #T_C(#i) -Z
‘GRP’ 22 T #Z_C(#i) #T_C(#i) -Z
‘GRP’ 23 T #Z_C(#i) #T_C(#i) -Z
ENDLOOP

This is the moment of positive temperature

This is the moment of negative temperature

I expect that the internal forces due to negative temperature are the opposite to those due to positive temperature, but the result are not as I expected.

I also did the analysis using LUSAS, the moment of negative temperature are the opposite to those due to positive temperature as shown below (LUSAS use N and m, so 436.4E3 on the beam means 436.4E3 Nm, which is equal to 436 KNm in Sofistik)

If you send the model, I can take a look at it.

Put all the files in one folder, run “Temp Compare.dat” in Teddy
Temp Compare.dat (135 Bytes)
Temp Compare-Aqua.dat (6.8 KB)
Temp Compare-ASE.dat (57 Bytes)
Temp Compare-SofiLoad.dat (1.1 KB)
Temp Compare-SofimshC.dat (19.7 KB)

Had a look at it. I have no idea what is wrong. Hope someone from Sofistik will answer. But they are probably on Christmas vacation.