Modelling of Viking Johnson

Couplings in CAESAR II

CAESAR

__________________________________

1998 describes Victaulic Coupling m odelling but

Jouko Tolonen
J Tolonen Services cc
www.jat.co.za

II docum entation

or

any known

publications that I know of do not describe how

to m odel VJ couplings. COADE newsletter June
this cannot be used for VJ couplings because of
the different coupling design philosophy.
This docum ent describes som e basic steps and

Introduction

m ethods how to m odel VJ couplings and flange

Viking Johnson (VJ, www.vikingjohnson.com )

couplings and flange adaptors are cost effective
solution for low pressure, low tem perature

adaptors and also gives som e ideas for line

design.

More

advanced

options

are

also

discussed.

applications where line flexibility is required. This

coupling technology is not new but I have found
that they are not fully understood. As a result
they are either not used or the lines are

Description of the couplings

Flange adaptor can be considered like half of a
VJ coupling. The following paragraphs refer to

incorrectly designed or erected.

flange adaptors. By reading the suppliers

Often low pressure lines where couplings are
used do not require very detailed design
calculations. Sam e applies for VJ couplings and
flange adaptors. Reading and following the
suppliers technical literature m ost applications
can be easily designed. However there are
applications where m ore detailed analysis is
beneficial. Typical such application is PF Pipes.
These are pipes that connect coal m ill to burners
at the boiler. Lines are low pressure lines but one
end of the line has large forced m ovem ent.
Typically som e 150 to 200 m m downwards and
40 to 100 m m horizontal. Operating tem perature
is both sides of 100 C.

Flange adaptors or

couplings can be used to com pensate these

m ovem ents.

Figure 1 QuickFit Flange Adaptor by Viking

and Johnson

In a low pressure low tem perature line stresses

are norm ally low and they are easy to calculate
or estim ate. Using VJ couplings and adaptors

technical literature it is easy to understand

couplings once you figure out the adaptors.

with correct design the stresses and also pipe

end

forces

can

be

substantially

reduced.

Traditional stress analysis is largely stress

Flange adaptor looks like a flange with enlarged

hub and som e extra bolts.

oriented. For a pipeline where VJ are used this is

not the case. Far m ore im portant are angular
m ovem ents and axial forces.

Flange adaptor cross sections shows the m ost

im portant design features.
These are:

___________________________________________________________________________
24 July 2008 Page 1 of 7

There is no m etal to metal connection

Manufacturers technical literature gives exact

between the pipe and the adaptor

details. Pressure thrust can be calculated using

There is a gap between the flange face

form ula:

and the pipe end

Connection between the flange adaptor

and the pipe is by a rubber seal, which is
com pressed by bolts and a loose ring.

Because of these design features flange adaptor

can rotate and allows sm all axial m ovem ent.
Rotation is lim ited to 3 and axial m ovem ent to 5
m m . Both are from neutral position. Larger axial

Instead of solid rods chains can be used also.

For a pipe designer it is im portant to understand
that it is not only pressure thrust that has to be
looked at. Com pression forces from external
sources or from therm al expansion have to be
looked at also. Calculated axial m ovem ent in
either direction m ust not exceed the 5 m m lim it!

Line design with large movements

VJ couplings and adaptors have lim ited axial
m ovem ent capacity. It is not sufficient to
com pensate large m ovem ents. It is a good
Figure 2 Cross section of a flange adaptor.

solution for instance for overland pipelines where

m ovem ent is possible but should not be allowed.

either am bient tem perature or fluid tem perature

Axial m ovem ent up to 5 m m is achieved by

changes require som e com pensation in axial

deform ation of the rubber seal. After that seal will

direction of the pipe. Such design can be

slide and m ost probably will get dam aged.

calculated

m anually

using

m anufacturers

literature.
No lateral m ovem ent is possible using one
For large m ovem ents different approach is

flange adaptor. Torsion is not allowed.

required. Instead of the axial com pensation the

line design is using the rotation possibility of the
flange adaptors. Connecting two flange adaptors
with a pipe between we have a cost effective
universal com pensator. For this to work we
need change of direction in our line.

Figure 3 Typical solution to carry pressure thrust

In the im age you can see four flange adaptors.

The high end of the line m oves down and left. In
addition there is norm al pipe expansion but it is

It is im portant to understand that flange adaptor

will not carry pressure thrust. It has to be carried
by additional devices or anchors in the line.
Anchors

have

to

be

in

regular

intervals.

___________________________________________________________________________
24 July 2008 Page 2 of 7

Figure 4 Typical PF line for a small power station. Left end of the line is the mill end. Flange adaptors are the round rings.
Note that there are no supports between the first and the fourth adaptor. These items must be free to move.

less than the forced m ovem ents. Downwards

not designed to carry axial load. Adaptors in the

m ovem ent is com pensated by the horizontal pair

vertical line m ust carry the pipe m ass below plus

of adaptors and the pipe between. Horizontal

the pressure thrust. These adaptors require

m ovem ents are sim ilarly com pensated by the

chains or rods to carry this load.

flange adaptors and pipe between in the vertical

section. How this works? Very sim ple. Flange

To increase the com pensation capacity cold pulls

adaptors are rotating and the pipe between is

can be used. It is generally beneficial that during

doing the sam e. Rotation is lim ited to 3 but

the operation load case flange adapters are near

increasing the length of the pipe between large

their neutral position. Erection tolerances and

m ovem ents are possible.

tem perature differences can be substantial.

Having the adaptors near their neutral position is

Face to Face

M aximum movement

good as they have their m axim um m ovem ent

1000

50

2000

100

3000

160

Modelling in CAESAR II

4000

210

So how to m odel the flange adaptor in CAESAR

5000

260

II? Sim ple, use expansion joint m odelling.

6000

310

7000

370

The m ost sim ple approach is to use gim bal

8000

420

com pensator m odelling. Referring to CAESAR II

9000

470

docum ent Application Guide Chapter 5 you

capacity still available.

m odel each flange adaptor individually. For this

1 Maximum movement in mm that pipe spool with two
flange adapters can compensate without cold pull.

m odelling CAESAR II requires five entries.

In the m ost sim plified case you enter high axial

Force required to rotate is low com pared to axial

stiffness. This will ignore any axial com pression

com pression of a flange adaptor. As a result

or extension of the adaptor. The error is m inim al

m inim al axial com pensation in the adaptors will

as the possible axial m ovem ent is sm all

happen.

com pared to other m ovem ents. If you do this you

have to check the axial forces at each adapter

You need to rem em ber that flange adaptors are

after calculation to m ake sure that you do not

___________________________________________________________________________
24 July 2008 Page 3 of 7

exceed the m axim um allowed force, which is the

Firstly we will use the finite length m odelling. Is

force when the seal slips. You do this by

this better is highly debatable. I discuss the

checking the global (or local) forces at the fitting.

problem s a little bit later. However there are

W e discuss later the m axim um allowed force

som e critical issues, which have to be noted if

and how to get it.

this m odelling is used.

As a translational stiffness you enter high value.

First issue is what length to use. I have seen 50

Lateral m ovem ent within single adaptor is not

to 60 m m used. I use 10 m m .

possible. If you like to use scientific approach

you can calculate radial stiffness of the pipe,

Secondly I will discuss the expansion joint entry

increase the calculated value by about 30% for

values.

the reinforcing effect of the adaptor and use this

value. After the calculation you should check the

In the sim ple m odelling we used high axial

lateral force and com pare it to m anufacturers

stiffness. W e can im prove this. It is im portant to

technical literature. They do not give m axim um

rem em ber

allowed force in Newton but you can calculate

capacity. Our design has to include suitable

this from their com m ents.

protection. Referring to our exam ple line in the

the

m axim um

axial

m ovem ent

vertical section we would use chains or rods to

Bending stiffness entry is the m ost im portant

carry the axial force. W e do not need to worry

value. W e discuss later how to get this value.

com pression here because there is no force to

After the calculation you need to check that in

lift the pipes. Fittings have to be free to rotate so

your design none of the adaptors have larger

we will use two chains. For practical reasons we

than 3 rotation in any of the load cases including

can assum e that they are not tight when they are

one after the cold pull.

erected.

As torsional stiffness you enter relatively high

To m odel our chains and adaptor stiffness we

value. Your design has to be such that the

use

torsion is elim inated as m uch as possible and

together with restraints. Firstly we enter axial

therefore this value is not critical. How high

stiffness of the adaptor into the expansion joint

torsion m om ent is actually allowed is not

axial stiffness entry. How to get this value we will

published by the m anufacturer.

discuss later.

Effective ID is the outside diam eter of the pipe.

Modelling of the chains is easy using the

CAESAR

II

expansion

joint

m odeller

restraint. W e use Y or +Y connected to the lower

This sim ple m odelling is fast to m odel but

end node of the finite length expansion joint and

requires substantial evaluation of the results to

Cnode to the upper node. Chain not being tight

m ake sure that the axial force or angular

we can enter 4 m m as a gap and then we can

m ovem ent lim its are not exceeded.

give som e stiffness value also. It is difficult to

design fully rigid chain with 2 pipe attachm ent

Complex modelling in CAESAR II

points. W e will use a generic 10 kN/m m

More com plex m odelling than the one described

stiffness.

above is possible in CAESAR II. I will discuss

som e of these possibilities using the pipe in the

In the horizontal pipe the m odelling is very sim ilar

figure as a exam ple.

to vertical. There are two issues you need to

___________________________________________________________________________
24 July 2008 Page 4 of 7

consider. First is do you have protection for axial

investigating this issue.

com pression in your design. If you use chains

the answer is no. In this case you need to check

To further im prove the m odel you could use

after calculation your global forces. Second is the

rotational restraint with 3 gap and stiffness.

direction of the restraint. You can give direction

However I am not recom m ending this. You have

to restraint if necessary, e.g. pipe axis is not in X,

to lim it the m axim um m ovem ent to 3. Exceeding

Y or Z.

this lim it m ay damage the flange adaptor. After

consultation with the m anufacturer you m ay

If you use rods with nuts on both sides of the

reconsider. If you enter stiffness for the restraint

fixing point you can m odel these sam e way as

the value will be less than your pipe bending

control rods. You m ay use gaps for the nuts.

stiffness, see cross section of the adapter.

Rem em ber to orient the rods correctly and do

Rotating pipe will connect inside the adaptor and

not use any RX, RZ or RZ restraints.

the ring at the end.

You have to enter translational stiffness. Use the

Torsional stiffness and effective diam eter entries

sam e value as you would use for sim ple

are sam e as for the sim ple m odelling.

m odelling. If you do not enter the value CAESAR

II tries to calculate it and it will be wrong.

Now we have still a problem what are the correct

axial and bending stiffness. W e look into this

CAESAR II docum entation indicates that if you

next.

enter translational stiffness you should not enter

the bending stiffness. Not in this case. If you do

VJ axial and bending spring rates

not enter value C AESAR II will calculate the

As you can see I use term s spring rates. Such

value using form ula:

term s are com m only used for flexible item s.

Manufacturer doesnt publish these values. They
depend on pipe diam eter, how tight the rubber
seal tightening bolts are and possibly into som e
extend what seal m aterial is used.

This gives a correct answer for m etal bellows but

The best m ethod to get these values is to test

not in this case. You have to enter correct

them . Tim e consum ing and expensive.

bending stiffness.
Viking and Johnson was very kind and gave m e
For the m etal bellows above form ula calculates

force that will result in seal slippage. Their tests

bending stiffness. Norm ally you have value for

show that on average 0.2T per pipe diam eter

bending flexibility. You can find explanation for

inch is the value to use. Using this value and NB

these two term s in COADE docum entation. The

400

difference between the two values is that the

0.2*1000*9.81 = 31 392 N or about 32 kN.

bending stiffness is four tim es the bending

Considering that the slippage happens after 5

flexibility. Calculation tests I have done (Version

m m m ovem ent we will get axial spring rate of

5.1 initial release) indicate that we need to enter

6400 N/m m . These values are m axim um values

bending flexibility. This is com m only known as

that can be achieved in test conditions. To be

bending

conservative I would not use the m axim um

spring

rate.

C O AD E

is

busy

pipe

(16")

we

can

calculate:

16

___________________________________________________________________________
24 July 2008 Page 5 of 7

values. To be conservative we could use for

The m ost im portant with these lines is to get the

instance as m axim um allowed axial force for

erection correct. W rongly erected lines will leak

NB400 pipe 16 kN and as axial spring rate of

and will not work. Erection m ust not be

3200 N/m m . This will result in larger axial

underestim ated. Especially if you specify cold

m ovem ent, which we have to check. It m ust be

pulls the erection becom es com plicated. Your

below 5 m m !

design calculations will help to establish correct

dim ensioning and if cold pulls are required.

Bending spring rate is m ore difficult. There are

no test results available. Such tests have to be

Calculation m ethod as described is accurate as

done sooner or later.

long as you can establish reasonably accurate

bending stiffness.

For the calculations we need a value, which has

correct m agnitude. Form ula given for m etal

W hat is the error and risk if the bending spring

bellows seem s to give this:

rate is wrong? It has sm all or no im pact on

calculated m ovem ents and generally very sm all
im pact on pipe stresses. Line stresses are
generally low and som e increase would not be
an issue. So what is left? Pipe connection forces
and m om ents to equipm ent.

Using NB400 pipe OD and axial spring rate 3200

N/m m we can calculate the bending spring rate

If you do not have tested bending spring rate the

(bending flexibility) as 1150 Nm /degr.

first is to use the above m ethod and then m ake

test calculations to establish what error in the

To check if the m agnitude is correct I used the

value would result in too high risk at the end

following:

connections. Based on the test calculations you

can then m ake a decision if tests are required.

Assum ing a pipeline is erected using 6 m long

NB400 pipes with 4.5 m m wall thickness would

Finite length modelling

pipe of its own m ass rotate after connected to

Finite length expansion joint m odelling system in

fixed flange adapter. The answer is yes. Now we

CAESAR II is designed for m etal bellows. It has

can calculate the m axim um bending m om ent at

built in form ulas, which do not necessarily work

the end of the pipe as 7870 Nm . Using 3

with VJ couplings or any other com pensating

m ovem ent our m axim um spring rate would be

elem ent that works differently to norm al m etal

2630 Nm /degr, which is sam e m agnitude as we

bellows.

calculated before.
First issue is that the program was designed so
Not the m ost scientific m ethod of establishing the

that you enter only bending or translational

im portant bending spring rate but it is best what

stiffness. The one that is left out is calculated

I can do without the tests.

using form ula applicable for m etal bellows. Only

in very rare cases you get correct results if you

Discussion

enter

Low tem perature low pressure lines where you

com pensator has im pact in these calculations.

would use flange adaptors and couplings are not

For define length com pensator you should

the critical lines to ASME B31.1 or ASME B31.3.

norm ally enter bending stiffness and not bending

only

one

stiffness.

Length

of

the

___________________________________________________________________________
24 July 2008 Page 6 of 7

flexibility (spring rate). In m y test calculations

som e results have been horribly wrong using
define length. Results include forces in wrong
direction, 180 degree rotation etc. The following
finite length com binations have worked:

short 10

mm

length

and

entering

bending flexibility

100 m m length, translational entry left

out and bending stiffness entered

In one specific case first option gave within 0.5%

sam e results as zero length option and the
second within 5% .
Zero length m odelling has always worked so I
recom m end to use it instead of finite length.

Conclusion
Using VJ flange adaptors is an effective m ethod
to com pensate large m ovem ents in low pressure
low

tem perature

lines.

CAESAR

II

gives

excellent possibilities to m odel them . You can

use very sim ple m odelling, which requires
detailed m anual evaluation of the results but you
can use also com prehensive m odelling, which
reduces the m anual checking.

___________________________________________________________________________
24 July 2008 Page 7 of 7