"COMPUTATIONAL DIFFICULTY OCCURRING DURING SEARCH AT CYCLE( ). THE CONSTRAINT WHICH CANNOT BE SATISFIED IS …
SUGGESTED STRATEGIES FOR RECOVERY.
1. RERUN WITH MORE FIXED VARIABLES, OR
2. LOWER THE UPPER LIMITS FOR CROSS SECTION VARIABLES, OR
3. CONVERT EFFORT TO RATE MODE.
RUN TERMINATED."
This is probably the most irksome message of all, and is difficult to prevent. It arises as the mathematical optimization process attempts to remedy a linear infeasibility in one of the cycle solutions it needs to solve on the way to the solution of the highly nonlinear optimization problem. In some cases it may occur because the user has used too narrow range limits for the design variables which are floated, and in other cases because they are too wide.
If the message occurs on the first design cycle, the problem could well be due to a weakness in the trial design generator required to launch the process of solution, assuming one has not otherwise over-constrained the design by an enthusiastic use of the USE command.
If the latter is not the case, you can contribute to improving the program by emailing your input to MDX. This will allow us to tune the trial design generator to at least handle the particular combination of data that triggered the problem.
You will need to find some combination of fixed and floating variable which will cooperate to produce convergence to a design. It is suggested, as a first strategy, you fix flange widths and web depth. In general, a good strategy is first to fix all plate widths and allow the program to determine thicknesses, or vice-versa. The difficulty seems to occur more often when both plate widths and thicknesses are floated, although this should not be taken as a generality.
"A CONSTRAINT IS VIOLIATED IN WHICH ALL VARIABLES HAVE BEEN FIXED. A RELAXATION IS NECESSARY. FIX ONE OF THESE AT A MORE CONSERVATIVE VALUE"
This can occur because of an over-enthusiastic use of range limit commands, particularly USE. It also could happen as the search process employs certain procedures to dampen nonlinear mathematical oscillations.
Follow the instruction issued by the program.
"DESIGN IS TOO CONFINED BY SPECIFIED LIMITS"
The user has placed range limits on certain variables, probably with the USE command, such that it is impossible to satisfy a constraint imposed by the specification. This message is followed by an indication of the nature of the constraint as an aid to removing the restriction.
Relax the range limits for variables likely to be involved in the constraint. For example, if the program indicates a constraint on shear stress is involved, the user might widen the range limits for web depth, web thickness, or stiffener spacing.
"NO ADEQUATE ROLLED SHAPE. THE FOLLOWING CONSTRAINT CANNOT BE SATISFIED"
ROLLED SHAPES has been requested with the CONDITION command, but no shape has been found which will carry the forces imposed on the beam.
If you believe there should be an adequate rolled shape, email your data to MDX.
"NOT ABLE TO CORRECT COMPUTATIONAL PROBLEM. RERUN WITH FIXED VALUES FOR SOME OF THE VARIABLES BEING ADJUSTED"
This problem is similar to the above, termination problem having the description "COMPUTATIONAL DIFFICULTY OCCURRING . . . ", except the problem occurs as the design is being rounded to commercially available values simultaneous with continuing optimization.
Relax the constraint by taking the action indicated. That is, if it appears the problem is with shear, fix the web thickness at a slightly larger value, or stiffener spacing at a smaller value.
"ONLY 19 VARIABLES ARE AVAILABLE FOR BEARING STIFFENERS AND TRANSVERSE STIFFENER SPACINGS PER WEB SECTION"
The user probably is requesting too many zones with PT data. Also a multispan continuous system may not have been subdivided into positive and negative moment regions by splices, in effect markedly increasing the number of stiffener spacing variables in the design problem to be solved.
Increase the number of web splices even though actual physical splices may not exist.
"PROBABLE OSCILLATION IN SEARCH PROCESS. SUGGEST RERUN WITH MORE FIXED VARABLES"
Discontinuities in AASHTO nonlinear constraints probably are causing an oscillation in the mathematical search process.
Fix more variables in data file and rerun.
"PROBLEM IS TOO CONSTRAINED FOR COMMERCIALLY AVAILABLE DESIGN VALUES. LIMITS NEED TO BE RELAXED FOR ONE OR MORE OF THE FOLLOWING DESIGN VARIABLES"
This is the same problem as indicated above ("DESIGN IS TOO CONFINED BY SPECIFIED LIMITS") except the difficulty occurs as rounding with continuing optimization is in progress, and the program is having difficulty finding a combination of commercially admissible plate thicknesses which will work.
Relax limits on indicated variables. It also might help to fix a plate thickness involved in the difficulty with the USE command.
"SYSTEM LIMIT REACHED ON NUMBER OF CONSTRAINTS GENERATED. ( ) GENERATED ( ) MAXIMUM. SUGGEST FEWER STIFFENER TRANSITIONS OR MORE SPLICES"
The number of constraints generated exceeds the size of the arrays necessary in solving the problem. If this is a persistent problem it might be possible for MDX to increase the size of the affected arrays.
Contact MDX for a program adjustment if the problem is persistent.