On-Line Detection of Pressure Relief

Device and Block Valve Leaks

Techniques using sealed radiation sources and unsealed radiotracers enhance the usefulness of
standard photographic X-ray technique for examining online industrial processes and equipment
thoroughly. This article focuses on radioisotope techniques that determine pressure relief device and
letdown block valve leaks online.

Dave Ferguson and Ed Wellman

ICI Tracerco, Houston, TX 77058

Introduction the fact that the amount of radiation that passes

through any material strictly depends on the density of
that material. This relationship is applied in practice

G
amma and X-radiations were put to practical
use almost as soon as they were discovered; by placing a sealed gamma-radiation source on one
within two months, X-rays were being used by side of a vessel and a sensitive radiation detector on
doctors to assist in surgical procedures. The industrial the opposite side.
use of radiation developed in the 1950s when universi- The source and detector are synchronously lowered
ty research laboratories and industrial research depart- down opposite sides of the vessel, measuring radiation
ments recognized the potential of radiation and the intensity as a function of column elevation and thus
cost of radioactive material became reasonable. generating a density profile of the operating vessel.
This paper deals with the use of sealed radiation Increased radiation transmission is seen in areas of
sources and unsealed radiotracers to more completely low material density inside the vessel and reduced
examine online industrial processes and equipment. transmission is seen in areas of high material density.
Interpretation of a distillation column density profile
Sealed Source Techniques provides information about the mechanical integrity of
the internals, the amount of liquid on the trays, and the
Column scanning presence of entrainment or foaming.

Sealed source techniques involve using a sealed Unsealed Source Techniques

source of radiation on the outside of a vessel or pipe.
The most commonly used sealed source technique is The use of unsealed sources of radiation allows
column scanning (Figure 1). This technique relies on direct measurement of the flow characteristics of

AMMONIA TECHNICAL MANUAL 48 1998

process fluid through a system. This is accomplished where problems exist with off-specification product.
by tagging the process stream with a chemically and Radiotracers are ideal tools for locating and confirm-
physically compatible radiotracer and following the ing leaks. Figure 3 shows a typical exchanger arrange-
labeled stream with external detectors positioned at ment. Radiotracer is injected into the high pressure
strategic locations. Radiotracers are available to tag feed to the exchangers. Detectors are placed on the
virtually any process stream in the liquid, solid, or high pressure inlets and outlets to monitor the passage
vapor phases. They also are not affected by the tem- of the tracer through that side. Detectors are also
perature, pressure, or chemical characteristics of the placed on the low pressure outlet lines to monitor for
native stream. the presence of tracer. Since the low pressure side is in
Radiotracers are regulated materials and can be han- series, a leak occurring in the top exchanger will cause
dled only by specially licensed agencies. Licensees are both leak detectors to respond. Suitable radiotracers
responsible for the material from cradle to grave. In are available for gas, hydrocarbon liquid, and aqueous
fact, even after the material is injected into the unit, it liquid streams.
remains the property of the group who injected it.
Only very small amounts of radiotracer materials are Identification Of Leaking Pressure Relief
consumed during a study, and the radiotracers chosen Devices and Block Valves
typically have a half-life measured in either minutes or
hours. It is the responsibility of the licensee to ensure Natural gas, synthesis gas, and ammonia are all sus-
that radioisotope disposal concentrations, whether in ceptible to loss through leaking pressure relief devices
the stream monitored or released into the environment, and letdown block valves. The Olin Ammonia
are below regulatory limits. Material at these concen- Production facility in Lake Charles, LA, was one of
trations is considered an exempted quantity, which the sites that used radioisotope techniques to identify
means that the level of radioactivity of the disposed leaking pressure relief devices and letdown block
material is in general no greater than the level of valves. On several occasions, the actual amount of
radioactivity that occurs naturally. ammonia produced compared to the usage of natural
gas led the plant operations personnel to suspect leaks.
Flow measurement Radiotracer diagnostic campaigns were mounted to
The most basic industrial use of a radiotracer is to identify the sources of the leaks.
measure the velocity of a fluid flowing in a pipe. On one occasion, a comprehensive investigation was
Figure 2 shows two detectors mounted onto a piece of launched in anticipation of an upcoming turnaround.
pipe. They are connected to rate meters and a computer. The normal turnaround procedure had been to remove,
A pulse of radiotracer is injected into the pipe through a dismantle, inspect, and rebuild all critical relief
bleeder valve and passes down the pipe at the speed of devices and a percentage of the less critical devices
the flowing fluid. As the pulse passes each detector, a and block valves. This procedure was costly and time
signal resembling a bell curve is generated on the consuming, and resulted in longer outages than would
computer. The computer calculates the time of flight otherwise have been required. Additionally, this proce-
of the pulse of radiotracer between the two detectors. dure only minimized, without eliminating, the risk that
The velocity of the fluid is calculated from the time of a leaking device would remain in service. Before this
flight and the measured distance between the detec- turnaround, the decision was made to use radiotracer
tors. Further calculations incorporating the cross-sec- technology to test each pressure relief device and let-
tional area of the pipe and the density of the fluid will down block valve while the unit was online. This
yield volumetric and mass-flow values, respectively. would ensure that only those devices and valves that
needed attention were placed on the turnaround main-
Leak detection tenance schedule.
Three different methods (direct leak testing, passing
Heat exchangers are the culprits in many situations leak testing, and additive flow measurement) were

AMMONIA TECHNICAL MANUAL 49 1998

 Demîster Pod
Clear Vapor
Liquid On Tray
Light Foam
Heavy Foam

Missing Troy
Tray Dabris

Flooded Tray
Liquid Weeping

Liquid Level

Daliclor
Detector D4
Response

Figure 1. Distillation column gamma scan. Figure 2. Flow measurement in a pipe.

used to test the possible leak points, based on the

f low te be type of access available.
Measured

Direct leak testing

The simplest method involves injecting radioactive

tracer into the line leading directly to the relief
device or block valve. Figure 4 is an illustration of
this technique. At the Olin facility, the natural gas
feed to the unit passed through the C-106 knockout
Computer Display drum. A relief device was connected to a spool piece
r that was mounted on a nozzle on top of the vessel.
This spool piece also had a sample connection.

p
« •U«
To test this relief device, detectors were mounted
on the line upstream and downstream of the device.
A small amount of tracer gas was injected into the
spool piece through the sample connection. As the
line tracer entered the spool piece, the detector upstream
MeaaVehtity* -f of the device responded to the presence of the tracer.
A few seconds later, the detector mounted down-
Figure 3. Typical arrangement of heat exchangers. stream of the relief device also responded. This
response indicated that tracer had leaked through the
device.

AMMONIA TECHNICAL MANUAL 50 1998

Passing leak testing the three-way valve to align the A valve. Neither of the
detectors responded to the second tracer injection.
Figure 5 illustrates the passing leak testing method.
This method is used when the line to which the relief Additive flow measurement
device or block valve is connected has no sample con-
nection. At the Olin facility, the relief devices on the In some cases, it is easiest to look for leaks by the
feed gas lines to the reformer were constructed in this additive flow measurement method. If several relief
manner. In this case, tracer gas was injected into the devices and/or block valve connections to the flare
main line, upstream of the spool piece to which the header are present, measuring the flow in different
relief device was mounted. If the relief device was segments of the header will identify the connection
leaking, some of the tracer gas would travel up the that is adding flow to the header.
spool piece and through the relief device as the bulk of Figure 6 shows some of the connections to the flare
the gas passed the connection of the spool piece. The header that serviced the back end of the Olin
detectors both upstream and downstream of the relief Ammonia facility. There was a suspected leak from
device would respond in the case of a leak. one of the relief devices of the synthesis gas compres-
As shown in Figure 5, the feed to the reformer was sors or the fuel gas compressor. An injection point
protected by two relief devices connected to a Y junc- existed near the front end of the header. One detector
tion with a three-way valve. Only one relief device was mounted on the header a few feet downstream of
was in service at a time. When the first test was per- the injection connection, and another detector was
formed, the system was relying on the B device for mounted on the header where each compressor's relief
protection. A response was recorded on both detectors, device was connected. Since there should have been
indicating a leak. The operations personnel then turned no flow at the front end of the header, a nitrogen hose

Detectors
Flare Line

Detectors

Injection

C-106
Knockout
Drum
Natural Gas to Reformer

Figure 4. Direct leak testing of pressure relief Figure 5. Passive leak testing of pressure relief
devices. devices.

AMMONIA TECHNICAL MANUAL 51 1998

 between detectors 2 and 3, indicating that the B synthesis
Injection
gas compressor relief device was not leaking. The flow
Nitrogen . between detectors 3 and 4, however, was measured at
about 5,000 std. ft3/h (40 L/s). This indicated a leak
through the fuel gas compressor relief device of about
"B" Syn Gas Compressor ^ 4,500 std. ftVh (36 L/s). The flow between detectors 4
3
and 5 remained about 5,000 std. fe/h (40 L/s), indicat-
ing that no leak existed from the A synthesis gas com-
Fuel Gas Compressor pressor.
During this turnaround preparation study at the Olin
facility, a total of 58 pressure relief devices and let-
down block valves were tested for leaks. A total of ten
leaks were found. During the turnaround, these were
all removed and repaired. Switching the Y valve on
the reformer relief device immediately eliminated one
leak source. The savings estimated by this simple
action paid for the week of work by the radiotracer
crew in 3 days.

Summary
Figure 6. Additive flow measurement of pressure
relief devices. As this case history illustrates, radioisotope tech-
niques can provide information that is either unat-
tainable or more expensive to attain by other
was connected to create a flow and keep the tracer gas means. This information is gathered online. Many
moving in the header. problems can be corrected or mitigated without a
When the tracer gas was injected, a flow rate was shut down. Even when a shut down is required, all
computed between detectors 1 and 2, 2 and 3, 3 and 4, preparations and purchases can be made to reduce
and 4 and 5. When the data were analyzed, the flow the amount of downtime and lost production. Either
due to the nitrogen (detectors 1 and 2) was measured way, these tools will allow engineers to improve
at about 500 std. ftVh (4 L/s). Flow did not change the profitability of their unit.

DISCUSSION
W. D. Verduijn, Kemira, Rozenberg: We suspect that increasing wall thickness the amount of radiotracer
we have a leak in the feed effluent exchanger of our that you use in an individual study also increases.
ammonia synthesis loop. Your radiotracer technique We frequently do testing all the way through the
seems to be an interesting option to shed some more loop. We've looked through converter shells that
light on that problem because we cannot really prove were four and five inches thick. Testing exchangers
the leak with ammonia concentration measurements, is usually considerably easier than that because you
but we very much suspect that there is one. However, can look at the pipe work on the inlet and outlet
the synthesis loop operates at a pressure of about 220 where the flow is all compressed to one location, so
bars, and the Unes around the exchanger have a very it wouldn't be difficult.
high wall thickness there. Would there be any upper P. Orphanides, Consultant: Concerning the last exam-
limit to the wall thickness of the lines that prohibits a ple with the relief valves, where you calculated differ-
successful application of your technique? ent flows going into the system having different mole-
Bucior: Yes, I'm sure there is some upper limit. It cular weights: Do you measure with the tracer volu-
would be a practical limit. Normally, to account for metric flows or mass flows, and how do you adjust for
AMMONIA TECHNICAL MANUAL 52 1998
the different densities of the fluids leaking into the to the cooling water system, is it possible, using your
nitrogen? method, to detect the leak, that is, which exchanger is
Bucior: Normally, what we do is measure the flow leaking?
velocity. It depends on how accurate you need the Bucior: Yes, because the radiotracer would cross
flow rate to be. You could estimate the density or you streams in strictly the same proportion as the process
could measure it using the gamma transmission tech- stream. You put the radiotracer in far enough
nique that was briefly discussed at the beginning of upstream of your detection locations to ensure that it's
the paper. There is a rig that allows measurement of thoroughly mixed across the flowing stream and,
the density of the flowing stream through the pipe if therefore, representative of the flowing stream. If the
you want to get a really accurate fluid density. The process gas is leaking, then the radiostaged process
bulk of the error in the technology comes from the gas is going to leak also, which enables detection of
manufacturers tolerating variations in the pipe diame- the source of the leak. I should probably point out that
ter when the pipe is constructed. The velocity mea- this particular measurement technique is good down
surements are accurate to within, certainly, a few hun- to about a 1% lower limit. To use external radiation
dredths of a second. The manufacturer's tolerances on detectors you need 1% of the high-pressure steam to
the ID of the pipe probably lead to an error somewhere leak into the low pressure stream. If further sensitivity
between 2 and 5% in the overall measurement. is required, you collect samples of the low-pressure
L. Zafar, Fauji Fertilizers, Pakistan: Can we use stream and assay them. With that technique, you can
radio isotopes for measuring levels in LNG tanks, get the sensitivity down to whatever you need it to be.
LPG tanks, and liquefied ammonia tanks? G. R. Prescott, Consultant: Did you say that you
Bucior: There's another technology I didn't actually could measure variation in wall thickness of a vessel
cover in this paper called neutron back scattering. It's or pipe using combination isotope and sensor as you
a device that consists of a source of high energy neu- move along?
trons collocated with a detector that's sensitive only to Bucior: No. What I was talking about was in answer
low energy neutrons. The neutrons give off their ener- to the question concerning measuring the density of
gies by collisions with hydrogen nuclei. You put this the flowing stream in the pipe. You could measure the
probe up against the side of a vessel, the fast neutrons thickness of the pipe, but ultrasonics would probably
go in, interact with the medium inside and, depending be quicker.
on how hydrogenous that material is, a number of Prescott: We have a special problem with multiwalled
them will be moderated to low energies, where they vessels in urea service where we breech an internal liner
bounce back out through the wall of the vessel. By and get corrosion behind it that is neither obvious nor
detecting how many neutrons bounce back, you can detectable. This corrosion is not amenable to ultrasound
get an indication of what the material is behind the technology. Would it be amenable to this process?
wall. The device is sensitive enough to tell the differ- Bucior: It could be, as one of the applications of
ence between various hydrogenous materials, so cer- the density measurements we use in ICI plants is
tainly the change from vapor phase to liquid phase measurement of the refractory thickness during
would be a remarkable difference in signal. That construction. Frequently we have different kinds of
would probably be the best route to go for measuring a refractories that are going to be laid in different
big vessel-liquid level. layers and as long as there's a density variation you
S. A. Al-Ghambi, Saudi Arabia: We have a closed can design a study to do whatever it is you need.
cooling water system. In the case of a gas leak going Prescott: Sounds like a good possibility.

AMMONIA TECHNICAL MANUAL 53 1998