Professional Documents
Culture Documents
This paper discusses the possibilities offered by ramp pressure loading tests to
generate data for determining the long term load-bearing characteristics of plastics
pipe systems. In addition to ramp pressure loading, procedures are also considered
where the pressure is held at a constant value for a period of time, after which the
product is subjected to a ramped pressure burst test. The method discussed relies on
the material in question obeying Miner’s law in its static fatigue behaviour. The pipe
system to which the procedure has been applied is Reinforced Thermoplastic Pipe
(RTP), in which the reinforcement is aramid fibre. However the results are expected
to apply to any polymeric system where there is evidence of conformity to Miner’s
law. A set of ramp loading tests, and constant pressure plus burst tests are reported
and compared with the results of conventional ‘constant pressure’ stress rupture tests
and a method is proposed for converting these results into ‘equivalent’ constant
pressure values. The results obtained on RTP using this new approach lie close to
those generated using constant pressure.
Introduction
Figure 2 shows the regression relationship, determined using constant pressure tests,
for one type of RTP, determined at 60ºC. The mean line and the 97.5% LPL are
shown. As in the majority of procedures for thermoplastics pipes1-3, the mean line was
determined by regression of log time on log pressure, rather than vice versa, because
this gives the most conservative prediction. Table 1 gives values of the regression line
parameters and the 20 year LPL calculated from these data, as well as those obtained
from other experiments using the novel loading profiles.
Procedures involving ramp loading, and employing Miner’s Law to interpret the
results, offer a solution to these problems. In addition, they offer the possibility of
deciding or determining when samples will fail during the test campaign. One variant
of the technique- constant pressure testing, followed by a ramped short term burst test
offers the most advantages. With this procedure it is possible to hold samples at
constant pressure in the conventional manner, then at fixed times carry out burst
testing.
The proposed procedures also have some drawbacks. For instance, Miner’s Law is an
empirical law and the product class in question needs to have been already shown to
obey it. It will also be seen that care is needed with timescales when converting ramp-
type results into equivalent constant pressure values.
Theory
The theory will be presented here in terms of pressure. However, it would be equally
valid in terms of hoop stress. Miner’s Law will be assumed to apply. This is an
empirical law and there is no reason why, a priori, a particular material system should
be expected to conform. Nevertheless, for a number of polymers the law has been
shown to be applicable with reasonable accuracy. Current evidence for aramid fibre
reinforced RTP5,6 suggests that it is obeyed. Miner’s Law states that, for a system
subject to i different stress values, (in this case, pressure, Pi) each lasting for duration,
ti, failure occurs when
ti
∑t =1 [1]
fi
where tfi is the value of the time to failure at pressure, Pi. If the constant pressure
stress rupture behaviour obeys
Pf = Ft f − G [2]
where Pf and tf are pressure and time to failure, respectively and F and G are
constants, then the Miner’s Law expression becomes
1
1
1 ∑ ti Pi G = 1 [3]
G
F
When pressure varies over time the summation can be replaced by an integral:
1
1
1 ∫ P G dt = 1 [4]
G
F
The following pressure histories will be considered here: (i) general pressure history;
(ii) pressure ramped linearly with time, and (iii) a constant pressure period, followed
by a ramped burst.
General pressure history. This would apply for any variation of pressure with time
during the test period. The pressure profile would need to be continuously recorded
until the sample failed. It can be seen from Equation [4] that
t =t f 1 1
∫
t =0
P G dt = F G [5]
As with constant pressure regression testing, several tests would be needed to enable
statistically accurate values of F and G to be determined. If a provisional value for G
were available, the above integral could be used to determine the equivalent value of
constant pressure, P*, that would cause failure in time, tf. Since P* = Ft −f G then
t =t f G
1 1
P* = ∫ P dt G
[6]
tf
t =0
Ramped pressure. With the pressure ramped linearly up to failure as in Figure 3 (so
that P = kt and, for failure, Pf = kt f ) then, for experiments at different ramp rates,
the Miner’s Law integral would become
1 1
1+ 1+
G G 1
1 1 1 t 1 t
∫t dt = k =k =F
1 2 G
k G G G G [7]
1 1 1
1 2
1
1 + 1 +
G G
so
1 1 1
1 1+ 1 1+
1
G
k t
1 1
G
=k t 2
G
2
G
= 1 + F G [8]
G
Taking into account the relationship between failure pressure and failure time, this
gives
1 1
1
PfGramp t f ramp = 1 + F G [9]
G
G
1
so Pf ramp = 1 + Ft −f ramp
G
[10]
G
This shows that the log-log relationship between failure pressure and time to failure in
ramp loading is very similar in form to the one for constant pressure testing. It is
possible, therefore, to produce a regression relationship in ramp loading in exactly the
same manner as for constant pressure testing. Comparing Equations [2] and [10] it can
be seen that the regression line slope, -G, is the same in each case. The other term
G
1
changes by a factor of 1 + . This provides a direct method of determining the
G
constant pressure regression constants from the results of ramp tests.
Figure 4 shows the results of the series of ramp loading tests carried out by
Wellstream on RTP at 60ºC. It can be seen that there is indeed a linear relationship
between failure pressure and failure time, and that the slope is fairly similar to the
value obtained in constant pressure testing (0.435, compared to 0.420). It is now
possible to calculate the equivalent pressure, P*, that would have been required to
cause the pipe to fail after a period, tf, under constant pressure.
G
G
P* = Pf [11]
1+ G
This pressure value can be used to determine a point, (tf, P*) for use in the constant
pressure regression relationship. Such points could be used alongside other points
determined from constant pressure tests, ramp loading tests, or both. Equation [11] of
course, requires a value for the exponent, G. Often a provisional value is known or an
initial value may be assumed. This results in a series of iterative calculations for the
power law parameters, which usually converge quite rapidly on steady values.
Constant pressure, followed by a ramped burst test from zero pressure. In this
case, following an extended period, t0 , at constant pressure, P0 , a burst test is
performed using a ramped pressure, as shown in Figure 3. Here the condition for
failure is
P = Pf 1 1
1
P0 t0 +
G
∫
P =0
P G dt = F G [12]
This leads to the following expression for the final failure pressure:
G
1 1 1
1+ G
Pf = k 1 + F G − t 0 P0G [13]
G
A useful application of this procedure would be to determine data points at long test
times in situations where the component had not failed under constant pressure. The
procedure can again be used to determine the ‘effective’ pressure value at which that
particular sample would have shown a time to failure of t0. Noting that P* = Ft0− G and
re-arranging Equation [13] gives
G
1 G 1+GG 1
P* = Pf + P0G [14]
kt0 1 + G
Again a provisional value of the regression line slope, G is needed, and this results in
an iterative calculation.
This leads to the following expression for the final failure pressure:
G
1+ 1 1
1 1
1+G
Pf = P0 G + k 1 + F G − t0 P0G [16]
G
Results.
Wellstream carried out a series of ramp loading pressure tests, the results of which
were reported to the JIP. Five of the sets of tests were carried out using ramp
conditions. The P* values were calculated using Equation [11]. An additional set
involved a 24 hour period of constant pressure, 130 bar, followed by a ramp up to
burst. For these data the P* values were calculated using Equation [17]. Figure 5
shows a regression plot based solely on the results of these tests. The group of points
at 24 hours were those from the constant pressure test, followed by burst. Table 1
shows that, although similar to the constant pressure data, these results, which cover a
shorter timescale, have a slightly steeper gradient and lower 20 year LPL.
Figure 6 shows a combination of both sets of results, which can be seen to fit very
well together Qualification procedures for thermoplastic pipe generally limit the data
points to times greater than 10 hours, so those ramp loading points at shorter times
were omitted when calculating the combined regression data shown in Table 1. The
regression constants and the LPL can be seen to be fairly similar to the constant
pressure values.
Concluding remarks.
This exercise suggests that ramp loading and similar tests could give results that are
compatible with those of constant pressure tests, and which could be suitable to be
used to determine data points for regression testing. The present results are based on
a limited set of data one just one system and it would be interesting for a more
comprehensive range of studies to be carried out to determine the validity of the
proposed procedures. The advantages offered include not only methods of dealing
with run-out samples at the end of long test runs but also the possibility of having
planned campaigns of testing where, for instance, groups of samples could be held
under constant pressure for a period, then ramp tested to burst all on the same day.
This would have advantages over the current procedure where such a group of
samples would fail at indeterminate intervals over a period of time.
Acknowledgements
References
3. ISO 9080: 2003, Thermoplastic pipes for the transport of fluids - methods of
extrapolation of hydrostatic stress rupture data to determine the long-term
hydrostatic strength of thermoplastic pipe materials.
4. ISO 14692-1, Petroleum and natural gas industries - glass reinforced (GRP) piping
- Parts 1-4.
F G 20 year LPL
[pressure [bar]
in bar]
PE liner PE cover
Aramid fibre reinforcement
200
Pressure, bar
150
100
50
10 100 1000 10000
Time to failure, hours
Figure 2. Model data set for modelling of RTP regression behaviour at 60ºC. The
lines shown are the mean (upper), the LPL (lower).
Ramp loading
Pf G
G
P* = Pf
1+ G
tf
Constant pressure, then burst test
Pf
G
1 G 1+GG 1
P* = f
P + P G
0
P0 kt0 1 + G
t0
Constant pressure, then ramp up to burst
Pf
G
1 G 1+G 1+ G
1
G
P* =
P G
− P G
+ P
P0 kt0 1 + G f 0 0
t0
Figure 3. Pressure test profiles, with expressions for the effective pressure for use in
regression analysis.
1000
100
0.1 1 10 100 1000
Time hours
Figure 4. Results of ramp loading tests on RTP carried out by Wellstream at 60ºC
and various pressurisation rates.
250
200
Pressure bar
150
100
50
0.1 1 10 100 1000
Time hours
200
Pressure bar
150
100
50
0.1 1 10 100 1000 10000
Time hours
Figure 6. Regression curve constructed from both constant pressure and ramp data
(60ºC). Open symbols: ramp data. Mean and LPL lines are shown.