Defining SN Curves Defining SN Curves

Before any fatigue can be computed, an SN curve must be defined. This is accomplished with the command:


     &REP_SELECT,  -OPTIONS

and the available options are:



     -SN, CURVE, TYPE, S(1), N(1), ..... S(n), N(n)

     -THICK_SN, TO, POWER, MAXCOR, FLAG

     -S_IMP_FACTOR, SFACTOR

     -F_STRESS, S_MULT

Once a curve has been defined, one need only specify

     -SN, CURVE

to select, CURVE, for use in computing cumulative damage ratios. MOSES has a set of curves automatically generated and hence these do not need to be defined to be used. The automatic ones are: the

For the WJ and CJ curves, the ones with "_G" include weld improvement due to grinding and the ones with "_P" include improvement due to hammer peening.

To define a curve, one first specifies TYPE which defines the type of curve. TYPE should be STRESS for a normal curve, or TENSION for a curve like the WIRE and CHAIN curves. The values S(i) and N(i) define the stress (or tension / break tension) and the corresponding number of cycles. After the curve is defined, one uses the -THICK_SN option to define a correction which depends on thickness. Most documents say that this is a reduction of the SN curve, but in MOSES, it is viewed as an increase in SCF. Thus, our factor is greater than one and is the inverse of what would be a reduction of SN. The only real difference is that POWER here is positive where with a reduction it is negative. Here, TO, is the nominal thickness (inches or mm) and POWER is the power of the correction; i.e. the curve will be corrected by a factor, FACT, where:

     FACT = MIN ( (THICK/TO)**POWER, MAXCOR )

and THICK is the thickness. The details here are governed by FLAG. If FLAG is USE_BRACE, then FACT computed as above with THICK equal to the brace thickness is used for both the brace and the chord. Otherwise FACT is computed for both the brace and the chord using their respective thicknesses. Here, MAXCOR can be omitted and it will be set to "infinity".

The -S_IMP_FACTOR option is used to define a "stress improvement factor". This factor simply reduces the SN curve by SFACTOR; i.e. the stress in the SN curve is divided by SFACTOR.

Neither the API X nor XP curves built into MOSES have any thickness correction! The reason is that there is quite a bit of ambiguity in API-RP2A about how to treat this question. In particular, depending on the details, no correction may be needed. If you do need it, however, it can be included with:

     &REP_SELECT  -SN X -THICK_SN  1 .25 1.5 USE_BRACE

This tells MOSES to use a correction on any brace having a thickness greater than 1 inch with a power of .25 up to a maximum factor of 1.5 and to use the brace factor for the chord. The 1.5 maximum comes from the fact that RP2A says that the X curve need not be reduced below the XP curve, and the X curve reduced by a factor of 1./1.5 has the same endurance limit as the XP curve.

The -F_STRESS option is a simple way of overcoming a deficiency in most of the published SN curves; they are only defined over a limited range of cycles. Most of the time, this does not matter, but the algorithm MOSES uses strictly considers only damage which occurs in the specified range. Without any further action, a beam with high stress might actually accumulate less damage if the stress were increased! This occurs because increasing the stress moves more of the harmful cycles outside the range of the SN curve. The value S_MULT is a multiplier which MOSES multiplies by the RMS of the stress spectrum to check for this bad behavior. If the product S_MULT * RHS is greater than the stress at the first point of the SN curve, then the stochastic integration is not performed, but damage is accumulated based on the first point in the SN curve. This is simply an estimate to keep one out of trouble. By default, S_MULT is 3.72, but you can set it to 0 to strictly obey the standard algorithm.