Tube rupture in a natural gas heater

Dynamic simulation supports the use of a pressure safety valve over a rupture
disk in the event of a tube rupture

Harry Z Ha and Patrick Stang Fluor Canada Ltd

A
fast pressure surge will occur during the shell side pressure. According to ASME Sec. VIII
tube rupture of a shell and tube Div. 1, a vessel has to be hydrotested at 130% of
exchanger, where a big pressure difference its maximum allowable working pressure.1 This
exists between the tube (high pressure) and shell would allow for design pressure excursion by
(low pressure) sides. The pressure rise can be so 130%, hence the 10/13th rule. Based on this
fast that a conventional pressure safety valve requirement, if the design pressure of the low
(PSV) may not be able to open properly to pressure side is inside the envelope of 10/13th of
accommodate the pressure surge. Dynamic the high pressure side, no extra protection on
simulation of tube ruptures in a shell and tube the low pressure side is warranted. For systems
exchanger generates the time related pressure where the low pressure side is below the 10/13th
and temperature profile for the exchanger when criteria, the tube rupture scenario can be miti-
the rupture happens. The results can provide a gated by increasing the design pressure of the
solid basis for process design, whereby whether low pressure exchanger side, and/or assuring
or not a rupture disk should be used instead of a that an open flow path can pass the tube rupture
PSV if the pressure rise is too quick for the PSV flow without exceeding the stipulated pressure,
to open. A dynamic simulation for tube rupture and/or providing pressure relief (API Std 5212).
analysis of a natural gas heater is conducted in However, for a heat exchanger with a large pres-
this study. The heater is a shell and tube sure difference between the high pressure and
exchanger, which utilises low pressure steam low pressure side, increasing the low pressure
(shell side) to heat the tube side high pressure side’s design pressure can be expensive and eval-
natural gas (113 barg). A detailed Hysys simula- uation of a tube rupture scenario is required.
tion model was built to reflect the external and When a tube rupture occurs within a heat
internal configurations of the exchanger. exchanger, high pressure fluid expands into the
A tube rupture scenario was set up by assum- low pressure side through the ruptured tube
ing one full bore rupture at the back side of the bore and creates a pressure surge on the low
tube sheet (per API Std 521 – 5.19.3). The pressure side. Depending on the nature of the
dynamic simulation model was validated with fluids on both sides, the pressure surge on the
the heat and material balance before running the low pressure side can be instantaneous. When
case scenarios. The pressure and temperature both tube and shell sides are in liquid service,
profile, as well as the relief load, were recorded the pressure surge is faster than the response
versus time during the event. The results support time for a PSV to open so that a rupture disk has
the process design of a spring loaded PSV as a to be installed.3 From an operation and mainte-
means of protection of heat exchanger shell side nance point of view, it is less desirable to use
instead of the rupture disk, in the case of tube rupture discs for a service since rupture disks
rupture of the natural gas heater. require more maintenance and the relief is also
Shell and tube exchangers are widely used in uncontrollable once it is broken. However, when
the oil, gas, and petrochemical industries. both shell and tube sides are in vapour service, a
Depending on the application, the tube side fluid PSV might have enough response time to react
pressure can be significantly different from the to the pressure surge caused by tube rupture,

www.digitalrefining.com/article/1001215 PTQ Q1 2016 1

 sure steam condenses on the
shell side at 3.5 barg and 147°C.
Flare
The exchanger operates under
LP steam no-flooding condition. The
3.5 barg PSV at condensate is drained to a
5.5 barg condensate pot and leaves the
system on level control (see
HP NG out Figure 1). The intended
exchanger configuration
includes a PSV (6R8) with a set
pressure of 5.5 barg installed on
the steam inlet line.
The BEU heat exchanger is
the ideal choice for heating low
fouling fluids like natural gas.
The U-Tube bundle is attached
to a single tube sheet, allowing
HP NG in the tubes to expand and
112.7 barg Condensate
contract freely under the influ-
LC ence of temperature variations.
The tube bundle is removable
and allows cleaning of the
outside surfaces of the tubes.
Lateral baffles are inserted to
Figure 1 Systematic sketch of a natural gas heater provide sufficient residence time
for segmental heating. The
considering the compressibility of vapour. steam flows horizontally from one segment to
This study will demonstrate such a case by the next. The studied exchanger has eight
dynamic simulation, where the high pressure baffles, which separate the shell side into nine
natural gas in the tube side is heated by low horizontal sections (see Figure 2).
pressure steam on the shell side. The dynamic The shell side of the exchanger is modelled as
simulation will capture the pressure rise profile nine separate vapour sections to match the
over time, which provides a solid basis to baffled shell segments of the exchanger design.
support the installation of a PSV These vapour sections are
instead of a rupture disk for a modelled as horizontal cylinders
tube rupture scenario in a natu- in the dynamic model. The
ral gas heater. segment length matches the
distance between two baffles,
System description and model and the segment diameter is
setup calculated based on the total
The system studied here is a shell side volume. The heating
typical natural gas heater in the steam is assumed to be evenly
inlet line of a LNG plant. The condensed as it passes through
heat exchanger is a TEMA BEU the nine baffled segments of the
type exchanger with high pres- shell. The steam may condense
sure natural gas passing through unevenly as it passes through
the tube side and low pressure each segment. However, the
steam condensing on the shell impact of condensing load
side. High pressure natural gas distribution on tube rupture
enters the tubes at 112.7 barg simulation is insignificant since
and 3.2°C, and leaves the Figure 2 3D depiction of a BEU the heat exchanger operates
exchanger at 80°C. Low pres- exchanger configuration under no-flooding condition.

2 PTQ Q1 2016 www.digitalrefining.com/article/1001215

The condensed steam is
7 50 000
removed from each segment by
45 000
using flow controllers to simu- 6
40 000
late the operation of the 35 000
5

Mass relief, kg/h

Pressure, barg
exchanger in a no-flood condi- 30 000
tion, with the condensate being 4 25 000
routed to the condensate pot. 3 20 000
A control valve set to 100% 15 000
open is used to simulate the 2 10 000
tube rupture event. The valve is 5000
1 PSV-F
sized to allow the maximum PSV-P 0
(choked) flow from the high 0 −5000
0 1 2 3 4 5 6 7 8 9 10
pressure tube side into the low Time, s
pressure shell side. The maxi-
mum flow is calculated from an Figure 3 Pressure profiles at the PSV inlet and the mass relief load versus time
assumed orifice. The orifice has
the equivalent open area that is 7 50 000
twice the internal cross-sec- 45 000
tional area of one tube to allow 6
40 000
for the entrance of the high 5 35 000

Mass relief, kg/h

Pressure, barg

pressure fluid through a single 30 000

4 25 000
tube rupture (per API STD 521
- 5.19.3). 3 20 000
15 000
2 10 000
Tube rupture dynamic runs and P-middle section
5000
simulation results 1 P-far point (PSV)
P-near rupture point 0
The dynamic simulation model
0 −5000
is tuned to match the steady 0 1 2 3 4 5 6 7 8 9 10
state simulation results before Time, s
running the tube rupture
scenario. Once the dynamic Figure 4 Shell side pressure profiles at different locations versus time
simulation reaches the steady
state condition and consistent results have been This is a conservative approach for PSV relief
established (key operating parameters become load calculation.
constant with time), an event scheduler is initi- The dynamic simulation integrates in one
ated to run the tube rupture case. The tube millisecond intervals to record the pressure
rupture case considers a sharp break in one waves. The simulation runs until a stable relief
tube and the tube rupture is assumed to occur load and relief pressure are reached. For this
at the backside of the tubesheet (API Std 521 system, the stable relief load occurs at approxi-
– 5.19.3). mately 10 seconds.
The following tube rupture assumptions are The pressure profiles at the PSV inlet, as well
programmed into the event scheduler: as the exchanger shell inlet, midpoint, and outlet
a) One tube ruptures at one second integration sections, are recorded with time to show the
time. pressure surge of the tube rupture event. The
b) All flow control valves for diverting segmental mass relief load is also recorded together with
flow stay in position when the tube rupture the physical properties required for the PSV
happens (assume the steam passing across the sizing. Figure 3 shows the pressure profile of the
shell side with the same flow pattern as normal). PSV inlet and the mass relief load as a function
c) Back flow of the high pressure natural gas to of time from the onset of the tube rupture event.
steam is not allowed when the shell pressure Figure 4 illustrates the pressure profiles at the
reaches the steam supply pressure and, there- exchanger shell inlet, middle section, and outlet
fore, steam flow is stopped at this condition. for the same tube rupture event.

www.digitalrefining.com/article/1001215 PTQ Q1 2016 3

 PSV performance by dynamic simulation an engineering margin (~10%)
should be taken between the
PSV performance Pressure Time, Rate, Note
exchanger shell side design
rise, bar sec bar/s pressure and the PSV set pres-
Dynamic pressure 2.0 0.40 5.0 Pressure rises rapidly with the PSV closed sure, to provide proper
rising rate to set pressure
Dynamic pressure rising 0.55 2.0 0.28 Pressure rising is mitigated once the PSV protection for shell side over
rate from set pressure to starts to open pressure.
fully open with 10% over-pressure
PSV open rate 0.55 0.1 5.5 Calculated from a typical PSV opening
time Conclusion
A dynamic simulation for the
Table 1 tube rupture analysis of a natu-
ral gas heater is conducted in
As Figures 3 and 4 show, tube rupture occurs this study. The heater is a shell and tube
at one second and the shell side pressure rises to exchanger, which utilises low pressure steam
5.5 barg set point in 0.4 seconds. The PSV starts (shell side) to heat high pressure natural gas
to open and the relief load increases gradually. (tube side – 112.7 barg). A detailed Hysys simu-
The PSV opens fully in two seconds and the lation model was built to represent the external
relief pressure and relief load have been stabi- and internal configurations of the exchanger.
lised. The recorded system pressure rising rate The tube rupture scenario was set up by assum-
(surge) is 5.0 bar/s before the PSV starts to ing one full bore rupture at the back side of the
open. The rising rate drops to 0.28 bar/s once tube sheet (per API Std 521 – 5.19.3). The pres-
the PSV opens (see Table 1). sure and temperature profile versus time, as well
The opening time for a spring-loaded valve is as the relief load, were recorded during the
between 50 and 100 milliseconds (API 521 HSE event. An immediate pressure surge was
Report-023, 2002).4 Single PSV installations, observed after the tube rupture. It takes approxi-
with non-fire relief scenarios, are designed to mately 0.4 seconds for the shell side pressure to
open fully with 10% over-pressure. The PSV reach the PSV set pressure. This pressure rise in
opening rate is defined as the over-pressure the shell of the natural gas heater needs to be
divided by 0.1 seconds (assuming it takes 100 maintained within the 10% over-pressure limit
milliseconds to open). allowed by code. The pressure response of this
For the tube rupture case in this study, the system, as determined by the dynamic simula-
PSV opening rate is 5.5 bar/s, which is greater tion, requires the PSV to fully open within
than the simulated pressure surge during the approximately two seconds to meet this crite-
tube rupture. The result supports the engi- rion. A typical spring-loaded PSV with an
neering option of using a PSV instead of a expected opening time of 100 milliseconds is
rupture disk for over-pressure protection of capable of meeting this requirement.
the heat exchanger under study. The study results support the process design: a
In the case studied here, the PSV is to be spring-loaded PSV instead of a rupture disk can
installed on the shell inlet piping, which is at the be used to protect the shell side of the exchanger
opposite end of the exchanger from the tube in case of tube rupture in the natural gas heater.
rupture. Depending on the piping arrangement
and the configuration of the shell and tube HYSYS is a mark of AspenTech.
exchanger, the pressure profile along the shell
paths can be different. As Figure 4 shows, the Reference
1 ASME Boiler and Pressure Vessel Code, Section VIII, Div. 1,
pressure profiles of the shell side near the
2008, UG 99(b).
rupture point (tube sheet), at the middle section,
2 API Standard 521, Pressure Relieving and Depressuring
and at the shell side inlet are slightly different.
Systems, 6th Ed, Jan 2014.
The pressure profile at the shell inlet (where the 3 API Standard 520, Sizing, Selection, and Installation of Pressure
PSV is installed) stays approximately 0.5 bar Relieving Devices in Refineries Part I – Sizing and Selection, 8th
lower than the pressure profile of the middle Ed, Dec 2008.
section and near the tube sheet, where the tube 4 API 521 HSE Report – Offshore Technology Report 023, Testing
rupture is assumed to occur. This suggests that and analysis of relief device opening times, prepared by PSI

4 PTQ Q1 2016 www.digitalrefining.com/article/1001215

(Pipeline Simulation and Integrity) Ltd for the Health and Safety the petroleum refining and petrochemicals industries, with an
Executive, UK, 65, 2002. emphasis on refinery processes. He holds a BSc degree from the
University of Alberta with a minor in advanced process control.
Harry Z Ha is a Principal Professional Engineer with Fluor
Canada Ltd. He has over 25 years of experience in R&D in the
petrochemical industries and has been working on oil sands LINKS
and heavy oil upgrading projects. In addition to process design,
he focuses on data and methods development to support More articles from: Fluor
process modelling and simulations, and holds a MS degree
in environmental engineering from Hong Kong University of More articles from the following categories:
Science and Technology and a PhD in chemical engineering from Combustion Systems and Engineering
the University of Alberta. Heat Transfer
Mechanical Engineering
Patrick Stang is a Process Engineering Director with Fluor
Process Modelling and Simulation
Canada Ltd with over 25 years of process design experience in

www.digitalrefining.com/article/1001215 PTQ Q1 2016 5