Fig. 21. Amplification factors and ill-posedness. (a) Test 8: amplification factors. (b) Test 9: amplification factors.of measurements are presented in Fig. 17. Amplification coefficientsand truncation orders are given in Fig. 18.Both tests 6 and 7 are simulated numerically using LAM3Hacquin (1996). The metal flow is constant in the roll gap, thus thestrip speed is higher at the exit of the contact length than at its entry.Thus, the relative speed between strip and roll and so the shearstress present a characteristic sign change in the contact. The neu-tral point is defined as the position where shear stress sign changesalong the contact length. When the exit tension increases (and theentry tension remains constant), there is a shift of the neutral pointtowards the entry of the contact. This usual behavior is due to thefact that the difference between the entry and exit applied forcescorresponds to the resultant tangential force in the contact lengththat decreases when the exit applied tension increases. Numericalsimulations of both tests 6 and 7 are compared with the experi-mental evaluation based on the present inverse method as shownin Fig. 19. Reasonable agreement is obtained for pressure, howeveras explained in the previous section, shear stress does not repro-duce the classical discontinuity at the neutral point. Even thoughthe neutral point shift (when the exit applied tension increases)is evaluated with the inverse method, contact shear stress is notsufficiently accurate to give a reliable experimental evidence. Asmentioned above, this limitation is mainly due to a too low trun-cation order (i.e. high regularization level of the inverse method).5.3. Influence of thickness reduction ratioTwo additional cold rolling tests are presented in order to showthe influence of the thickness reduction ratio (from 10% for test8–80% for test 9). Thin strips are rolled alike tests 6 and 7 and rollingparameters are presented in Tables 8 and 9. Unfortunately, thesetests have been performed with the acquisition station at 960 Hz.Thus, for test 8 with short contact length (6.25 mm) very few mea-surement points are recorded in the contact vicinity as shown inFig.