When you perform a fatigue-crack propagation test under load control with most, if not all, standard specimen geometries by maintaining a constant alternating load, you will be growing the crack at an increasing ΔK, although you should maintain the same load ratio (R = Pmin/Pmax).  This is a sound means to measure the da/dN vs. ΔK curve at intermediate to high growth rates (by moving up the curve - so to speak).  However, if you wish to measure lower growth rates that approach the fatigue threshold, i.e., below ~10-9 m/cycle, it is generally better, and the ASTM Standard E647 requires it, to grow the crack at decreasing growth rates (by moving down the curve) until the crack nominally "arrests" at the fatigue threshold.  This is best done by programmed load shedding (at constant R, of course) with a so-called normalized K-gradient of 1/K(dK/da) of typically -0.08 mm-1 or so.  To measure very high load ratio thresholds, increasing R-ratio loading can be used with a programmed constant Kmax-increasing Kmin spectrum.  This can give values of the fatigue threshold at very high R-ratios well above 0.9, but it's not such a good means to measure growth rates as the R-ratio is no longer constant (see Boyce, Ritchie, Effect of load ratio and maximum stress intensity on the fatigue threshold in Ti-6Al-4V, Eng. Fract. Mech., vol. 68, 2001, pp. 129-47).

However, depending what you require, I would refer to the latest version of the ASTM Standard E647 for the full details.

I'd only like to add that while constant ΔP FCG testing is the main recommended way of doing an increasing ΔK FCG in ASTM Standard E647 for da/dN > 10^-8 m/cycle, it is not the only way.  Alternatively, one can use a constant positive K-gradient and this will give the advantage that the rate of increasing ΔK is specimen independent.  In contrast, when doing constant ΔP FCG testing the rate of increasing ΔK corresponds to the relation between K and crack length, a, for that specific sample geometry. 

No - due to the test control mechanism. In the first case DK increases in accordance with the material properties in the second case in a manner controlled by the operator.

I disagree with that last comment.  For constant ΔP FCG testing, the rate of change of delta K during the test will depend on the selected sample geometry and dimensions regardless of the material properties.  For example, a W = 25 mm C(T) sample will have a different K-gradient than a W = 50 mm C(T) or a W = 25 mm M(T), and in no case will the K-gradient be constant but rather it will change during the test. So the operator selects the  K-gradient indirectly by the specimen selection.  In contrast, the operator can control the K-gradient to be constant and the same among different sample types by setting a positive, constant K-gradient.  See section 8.7 of ASTM Standard E647 for more details. 

Sir, i am using MTS machine. It has both options to peform FCG. First one is Delta K control which includes K increasing and decreasing test depending on normalized K gradient value (C=1/k) either positive or negative for increasing or decreasing K. Second one is force control. Pls clear following doubts.

1. Usually tests are performed in which mode delta k increasing or constant load?

2. what is difference between delta k increasing and constant load test?

3 when to use k decreasing test?