BULLETIN POLICY.
The COADE Mechanical Engineering
News Bulletin is published on a
monthly basis from the COADE/
McGraw-Hill offices in Houston,
Texas. The Bulletin is intended to
provide information about software
applications and development for Me-
chanical Engineers serving the power,
petrochemical and related industries.
Additionally the Bulletin will function
as the official notification vehicle for
software errors discovered in those
mechanical programs offered by
COADE/McGraw-Hill.
PC HARDWARE FOR THE
ENGINEERING USER (PART 4)
The last issue of "Mechanical Engi-
neering News" discussed a few of
the unknowns about OS/2, the new
operating system from Microsoft and
IBM. Since the last issue, only a few
new conclusions can be drawn:
CONTENTS IN THIS ISSUE
PC HARDWARE FOR THE ENGINEERING USER arr 4
MEMORY REQUIREMENTS
MACHINE TIMES.
MORE SPEED.
SUSTAINED & EXPANSION STRESS CASES.
CAESAR Il, NEW VERSIONS,
DYNAMICS ..
COADE/McGrawHill
TNA
A minimum configuration will
require at least 2 Mbytes of
memory. This does NOT include
the memory required by the
application package.
There are at this time few de-
velopment aids for writers of
engineering programs. Porting
engineering software to the
OS/2 environment is a
particular unknown. OS/2 still
remains targeted for the much
larger business PC market.
Current literature indicates that
only 15% of the developer base
is contemplating the OS/2
transition. Most developers are
waiting for OS/3, since there
are limitations with OS/2
involving the nature of the 80286
chip. It has been written that a
truly useful, full functioned
multi-tasking DOS type PC
environment is still at least 5
years away.
We will continue to monitor, and re-
port on, developments in this area,
eLkibnnMORE SPEED (Cont'd)
other would render some programs useless. Practically, the user must buy and in-
stall both the coprocessor board and the 80387 math chip. This brings the cost for
a complete coprocessor system to around $2,700. Unless there is an overwhelming
interest by the users in the accelerator boards, COADE will not attempt to offer
software specially prepared for these add-on boards.
SUSTAINED & EXPANSION STRESS CASES
We have had several requests to discuss in greater detail the sustained and ex-
pansion stress calculations associated with piping code compliance. Many engi-
neers are not aware of what comprises these load cases or how they are obtained.
Support calls are still questioning how a +¥ restraint can appear to hold the pipe
down, (in the expansion case). The following paragraphs attempt to explain these
items.
To understand the load case methodology we must understand first, what the
piping codes requite and second, how the analysis can be set up to satisfy these
requirements, ‘The answer to the first of these questions comes directly from the
B31.3 piping code:
Paragraph 319.23 of B31.3 defines the "displacement stress range” as the differ-
ence between the extreme displacement condition and the original (as-installed)
condition.
From an analysis perspective, we must compute an operating case which repre-
sents the "extreme displacement condition", and an as-installed case, and then find
the displacement difference between them.
This is exactly how CAESAR recommends the expansion stress case:
where D1 represents the operating displacements and D2 represents the as-in-
stalled displacements.
For linear piping systems, i.e. systems without single directional restraints, gaps or
friction, this is equivalvent to performing a thermal only run, For nonlinear sys-
tems the two separate load cases, (operating and as-installed) must be run sepa-
rately and the difference taken.
CAESAR terms the as-installed load case the SUSTAINED case. CAESAR'S recom-
(cont'd on page 5)
e‘SUSTAINED & EXPANSION STRESS CASES
mendations for a job normally appear
- (OPERATING) T+P+W
- (SUSTAINED) P+W
- (EXPANSION) D1 - D2
Where the third load case: D1- D2, is the difference between the first and
second load cases.
It is this third load case that answers the question:
How can a +Y restraint be holding the pipe down
The answer is that the expansion case represents a “difference” or a "range" of
values. If the load on a +Y restraint is 100 tb. in the operating case and 200 Ib.
in the sustained case, the “range” of the loads on the restraint is -100 Ib. At no
time however between the operating and the sustained conditions does the +Y
restraint hold the pipe down !!
Much of the confusion here is due to the way older pipe stress programs
computed the expansion stress. These programs were originally linear, and as
such always used the THERMAL ONLY run to compute the expansion stress. i.e.
WHEN A PROBLEM IS LINEAR: D1 - D2 = THERMAL ONLY CASE
When these older programs included nonlinear effects there was no convenient
way to add displacement range calculations without restructuring the entire
output processing scheme. At the time the best way around this problem was
to!
Run an operating case to determine the nonlinear restraint
configuration for all restraints in the problem.
Fix all of the restraints as per this operating configuration, i.c. if
a 1-D restraint lifted off then take it out of the problem for
all subsequent cases.
Run a thermal only run as an approximation to the expansion
stress range.
Run a weight + pressure run as an approximation to the
as-installed, or sustained case.
This was a reasonable solution for the time, but omitted certain components of
the stresses from the expansion case. The simple job below illustrates the
problems (Cont'd on page 6)SUSTAINED _& EXPANSION STRESS CASES (Cont'd)
that arise with this type of solution:
‘Thema Growth up = 15 In
oO)
Renrsice ge = 17001 SS singe irectiont
Mope = 1235. ‘Resaint Lifted Off
‘ORERATING OSITION
SINSTALLED POSITION
When the vessel heats up it causes the node 10 to lift off of its single directional
support. In this position there is a much greater bending moment at node 5.
The stress range due to thermal expansion is calculated as:
EXPANSION STRESS
( Mope - Minstalled ) / Z
(1235 - 97.) 1 Z
1138 1 Z
where:
Mope is the operating moment at node 5,
Minstalled is the installed moment at node 5, and
Z_ is the pipe section modulus.
‘The as-installed (or SUSTAINED) stress is computed from:
SUSTAINED STRESS
Minstalled / Z
97 1Z
Older programs would handle this job:
(Cont'd on page 7)
a TTSUSTAINED & EXPANSION STRESS CASES (Cont'd)
Run the operating case and determine that node 10 has
lifted off.
~ Take the restraint at node 10 out of the model.
~ Run the weight case without node 10.
- Run a thermal only case without node 10.
The expansion stress in the older programs is calculated from the thermal only
case run in step 4 above. A thermal only run for the geometry above shows zero
moments at node 5.
EXPANSION STRESS (Older Programs)
0.0/Z
The sustained stress for these programs is found from the weight case without
node 10:
SUSTAINED STRESS (Older Programs)
= Mope/Z
1235 /Z
Older programs omit the displacement range component from the expansion stress
and add it to the sustained stress. Newer programs making the exact
displacement range calculation add the displacement range component to the
expansion stress (where it belongs), and leave it as an option to the user to add it
(conservatively) to the sustained stress or not.
Mr. Tony Paulin (the author of CAESAR II) presented a paper on the mathematical
formulation of this problem at the Chicago PVP meeting. User's can call COADE to
receive a copy of Mr. Paulin's paper.
CAESAR II, NEW_VERSIONS.
Certain of the version 2.2 items required more extensive quality assurance than
(Cont'd on page 8)