The best measurement technique will depend in what conditions do you want measure it. Type of lubrication, load, speed, type of contact, etc. Please be more specific since CoF is a characteristic of a process rather than materials. It will be better if you select the test conditions close to those used in practice for specific components.
U can use a fixture to apply a constant pressure on two steel strips, and then apply a tensile load to the end of strips. Then, calculate the Coef, or fric. from the known relation F=mN
It is important to note that often obtaining accurate and repeatable friction coefficient measurements/results is a difficult challenge. This is the case even in controlled laboratory settings.
I understand you want to measure the dry sliding friction coefficient. This means, without any lubrication of the surfaces. The problem is not simple. There is a static friction coefficient which is in general (e.g. for steel) larger than the kinetic coeffcient, which depends also on the relative velocity of the two surfaces.
For simple, classroom applications see:
http://www.tribology-abc.com/abc/friction.htm
For engineering applications see:
ASTM G115 - 10
Standard Guide for Measuring and Reporting Friction Coefficients (http://www.astm.org/Standards/G115.htm)
I am surprised that Valéo France does not have data on this subject. You should ask your central for some input, have you ?.
Your case is very complex since in a forming tool the friction conditions vary from point to point pressure not being allover the same and relative displacement, even at constant tool speed, being variable as well since stretching is variable. I think you should generate same conditions as in a tool and make a simplified model with which you could extrapolate obtained values from the research model conditions to the real tool. Under same conditions I understand the same relative geometry (for instance same ratios tool radius to sheet thickness and same gaps between tool parts), same lubricants since lubricant composition is determinant to its adherence to sliding surfaces and as far as possible same sliding conditions as speed. I would make a tool where I could deform a rectangular sheet to form a "U". The same pressure should be applied on the sheet as in real tools. this will simplify the model all being plane. With sheets of different width you could determine the proportionality between measured forces and width. By changing internal tool dimensions you could change gap conditions and thus stretching and sliding conditions. You could also change the speed and so on.
It is as I wrote a complex and long way till you will have usable data.
As a comment to Mr. Predoi, in metal sheet forming friction is never dry and pressure are such that the weight method is not practical.