DESIGNING ON-LINE CHROMATOGRAPH SYSTEMS

FOR LIQUID FRACTIONATION FACILITIES

Murray Fraser
Daniel Measurement and Control
Houston, Texas
Copyright 1998 Daniel Measurement and Control
Filename: gpa-98.wpd

ABSTRACT

Liquid fractionation plants can optimize their operations by installing on-line gas
chromatograph systems that have been properly designed to provide reliable, fast,
accurate results.
One of the most important, but overlooked, facets of designing an on-line analytical
system for gas processing facilities is the sample conditioning system (SCS). The
sample delivered to the Gas Chromatograph (GC) must truly represent the process
media if the measurement is to be accurate — or even meaningful. Samples may
be transported to the GC in either gas or liquid phase, but they will ultimately be
analyzed in the gas phase only. Selection of sample location and careful attention
to sample phase (liquid or gas ) is required to ensure optimum system performance.

Discussed in this paper are: typical process conditions, GC location, sample

transport systems, and details of both vapor- and liquid-phase sample conditioning.

INTRODUCTION

The SCS , purchased from the GC vendor or separately along with an optional
analyzer shelter and associated instrumentation, performs three functions:
1- Sample transport system,
2- Sample conditioning and stream switching, and
3- Sample injection.

All three parts must operate together to supply a clean, representative, single-
phase sample to the analyzer. Each measurement stream has process-specific
requirements centered around the need to transport and maintain a single-phase
s a m p l e to the sample inject valve located in or near the analyzer oven. After
sample injection, the sample is mixed with a light carrier gas and carried through the
columns in the vapor phase for component separation and subsequent analysis.
On-line gas chromatographs must provide fast, accurate, reliable results for any
process control system to fully optimize plant performance. Key factors are:
1- Reliability,
2 -Analytical Speed, and
3- Cost.

1
THE PROCESS

Liquid fractionation towers (columns) can be controlled by optimizing the inlet feed
fl o w rate, reflux flow rate, reflux temperature and column pressure. On-line
chromatographs can provide measured results of both the composition in column
overheads (gas phase ), column bottoms and inlets (liquid phase) for process
control. With GC data, changes in composition at the inlet to each column can be
compensated for as part of the overall control scheme. Tighter overhead product
specifications can then be met and reduced operating costs can result. ( For a
more complete description of the process, refer to GPA Engineering Data Book
Volume II, Section 19).

The following process conditions, typical of an NGL fractionation facility, serve as

an example for design of an analytical system as part of a larger process control
system.

Figure 1
2
 Table 1
Typical Stream Compositions (mole%)
Component Inlet De-eth De-prop De-but De-eth De-prop De-but
(mole%) feed over-hd over-hd over-hd bottom bottom bottom
Nitrogen Trace Trace 0 0 0 0 0
Methane 0-3% 0-2% 0 0 0 0 0
C02 0-3% 0-2% Trace 0 0 0 0
Ethane 0-35% 0-100% 0-10% Trace 0-10% Trace 0
Propane 0-35% 0-20% 0-100% 0-5% 0-60% 0-5% Trace
Iso-Butane 0-15% 0-2% 0-5% 0-50% 0-20% 0-30% 0-2%
N-Butane 0-15% Trace trace 0-70% 0-40% 0-50% 0-15%
Iso-Pentane 0-5% 0 0 0-2% 0-5% 0-10% 0-30%
N-Pentane 0-5% 0 0 0-2% 0-5% 0-10% 0-35%
Hexanes + 0-3% 0 0 Trace 0-10% 0-40% 0-55%

Table 2
Typical Process Conditions
Process Inlet De-eth De-prop De-but De-eth De-prop De-but
Conditions feed over-hd over-hd over-hd bottom bottom Bottom
Pressure at 400 30 psig 15 psig 20 psig 400 260 psig 100 psig
sample point psig psig
Temperatur 70 F 95 F 80 F 130 F 220 F 260 F 200 F
e at sample
point
Phase at Liquid Vapor Vapor Vapor Liquid Liquid Liquid
sample point
Phase to Vapor Vapor Vapor Vapor Liquid Liquid Liquid
sample
system oven
Phase to Vapor Vapor Vapor Vapor Vapor Liquid Liquid
analyzer
oven

3
THE GAS CHROMATOGRAPH

This paper does not attempt to discuss the merits of one application method over
a n o th er or the differences in design from one manufacturer to another. GC
components such as valves, columns, detectors, electronic hardware and software
can be configured in a multitude of ways for the same application. When evaluating
various methods, these quality criteria should be considered:
1- Analytical speed (the time it takes to transport, separate and measure each
component of interest),
2- Complete baseline separation of all measured components from one
another on a chromatogram, and
3- Suitability for packaging in an analyzer building, three sided shelter or
outdoors.

GC Location

The GC should be located as close to the sample point as possible to allow for short
sample transport lines. This is frequently a compromise between location of the
analyzer building and available site space to locate the building. GC location will
determine the sample system lag time which can account for the majority of the
overall system lag time.

There is a trade-off between cost and analytical speed when considering the location
of a common analyzer building. Sample transport time (lag time) can introduce a
large delay into the response of the analytical system. The best solution for
minimizing sample lag time is to reduce the volume of sample in the transport lines
and the sample conditioning system. By minimizing the volume, the velocity at which
the sample travels can be increased. Shorter sample lines and smaller diameter
tubing (1/8 inch from 1/4 inch) both help to reduce sample volumes.

Vaporizing liquid samples will reduce the sample volume and increase the sample
velocity a great deal. Once a liquid or heavy vapor phase sample has been
vaporized, the sample transport tubing must be heat-traced. Heat tracing can be
costlywhen multiple sample lines travel several hundred feet. Installation costs for
liquid sample lines are substantially less than for heat-traced vapor phase sample
lines.
The sample return flow rate can also be set in the SCS to increase sample flow on
either liquid or gas phase sample lines depending on the suitability of returning
h i g h e r volumes of sample to the process return point. For example, it may be
considered wasteful to return high volumes of on-spec products to the flare stack.

An alternative solution is to locate the GCs outdoors close to the sample point rather
than in a central analyzer building.

4
This can greatly reduce the installed cost and increase the speed of the overall
system, therebyimproving plant control system performance. There is considerable
disagreement within industry about the merits of locating GCs outdoors.

The perceived disadvantages with outdoor installations include the following:

1- Difficult to service in adverse weather conditions,
2 - R e duced operational life associated with exposing electronics to the
outside environment
3- Measurement stability.

Apartial solution is to locate the GC controller, with the majority of the electronics,
in a control room environment. Technicians can then perform maintenance work
w i th PC-based software from the control room or offsite via modem thereby
minimizing technician time at the actual GC.

Communications

In recent years there has been a trend in the process industry toward open protocols
a n d open systems as opposed to proprietary data highways. Technology is
changing rapidlyand the ability to keep pace and support proprietary data highways
is increasinglydifficult. Several well-established protocols are fully supported by the
control systems vendors for serial interfacing to DCSs and PLCs. These interfaces
are usually designed around a dedicated protocol converter often referred to as a
gateway. These gateways plug into a DCS or PLC bus and provide a serial interface
to other field devices using the same protocol. As a minimum, the GC should have
conventional protocol support on redundant serial ports. There is no real need for
proprietary high speed data highways to pass GC data that does not change for
several minutes. A serial interface is more than adequate for on-line readings as
well as full diagnostics, chromatograms and other troubleshooting tools.

THE SAMPLE CONDITIONING SYSTEM

Basic Rules of Sample System Design

The first rule in sample system design is to ensure that sample probes extract a
sample from a point 1/3 to 2/3 of the distance into the process piping. This will help
ensure a representative sample with minimal contaminants from piping internal
walls.

In general, the required single-phase samples are maintained by temperature (heat)

on vapor streams and pressure on liquid streams. Vaporization can occur at several
points depending on the composition and the practicality in keeping the sample
completely in a single phase from sample point to sample valve.

5
Vapor Phase Samples

The Sample Transport System

The compositions listed in Table 1 represent the highest full-scale range for each
component; typical operating ranges may be less. To determine the heat-tracing
requirements, the full- scale range of the heaviest components must be used. This
e n sures that a representative sample will be delivered under all process and
environmental conditions. The temperature, pressure and composition at the
sample point will determine the initial sample phase. The same principles hold true
for the sample transport system and the sample conditioning system.

Vapor Phase Sample Inject

As noted, all streams must be in the vapor phase after the sample valve for GC
analysis. Table 2 shows that the inlet feed and dethanizer bottoms are sampled in
the liquid-phase.

Figure 2 shows the inlet stream is vaporized at the sample point and transported to
the GC in the vapor-phase. The dethanizer bottoms stream is sampled and
tra n s p o rte d to the GC in the liquid phase and vaporized in the GC sample
conditioning system. In either case, the vaporizing pressure regulator (VPR) adds
heat at the exact point where the sample pressure is dropped to counteract the
cooling effect that occurs when pressure is reduced (Joules Thompson effect). The
g re a te r th e pressure drop, the greater the cooling and the greater the heat
requirement to maintain complete vaporization of the heavier components. The
remaining two streams (dethanizer overhead & depropanizer overhead) are vapor-
phase at the sample point with heat tracing installed to simply maintain the sample
in the vapor phase. Additional pressure reduction occurs in the sample system oven
which operates at an elevated temperature (170
F/70
C in this case).

Vapor-phase samples should be injected at atmospheric pressure to increase the

repeatabilityof the measured sample volume. To accomplish this, a sample shut off
valve (SSO), located in the analyzer oven, typically shuts off sample flow about five
seconds prior to sample injection. This references the sample loop to atmospheric
pressure to improve the repeatability of the sample volume injected. The sample
ve n t a nd measure vents from the analyzer should be returned to atmospheric
pressure.

Sample Conditioning

To ensure a single phase for vapor streams in figure 2, the SCS should increase
sample temperature as the sample travels closer to the sample valve.

6
For instance, sample temperature should increase by 5
C (after pressure regulation)
from the sample transport system to the sample conditioning system and again from
th e sample conditioning system to the analyzer oven. A vaporizing pressure
regulator (VPR) can be used at the sample point to assist in temperature control
(refer to the inlet stream on figure 2). The remaining streams (dethanizer overhead,
dethanizer bottoms and depropanizer overheads) are already in the vapor phase at
the sample point. Pressure regulation on these streams does not occur until the
sample has traveled into the SCS oven where adequate heat is available. The low
s a m p l e p re ssures will be required to transport the samples to the sample
conditioning system where pressure will be regulated. Depending on the pressure
loss over the distance traveled, there may be very little sample pressure available
at the sample system. In most cases 2-5 psig of sample pressure is adequate.
Individual stream pressures should be set for balanced sample flow, as indicated by
the sample flow meter prior to sample vent two. The stream to stream purge is set
from the lower solenoid purge flow to eliminate cross stream contamination.

Sample transport lines require heat tracing from 55
C - 70
C (120
F - 170
F).
When installing heat-traced sample lines it is important to ensure there are no cold
spots after the sample probe or VPR. During winter conditions a few inches of cold
tubing can be enough to cool samples such as these enough to drop out heavier
components into the liquid phase. The filters shown on streams one, two, three and
four in figure 2 can be a coalescing type or a membrane type. Either type of filtration
method is designed to remove solids and liquids while allowing vapor phase
samples to proceed. Membrane filters offer superior liquids removal in saturated
conditions without allowing downstream carryover. Once saturated, coalescing filters
allow liquids to pass. Coalescing filters offer a more rugged solution for streams
containing excessive particulate.

The process conditions at the sample point may vary from the conditions listed in
table 1 due to access to process piping for installation of the sample probe. The
closer the sample probe can be located to the tower overheads and bottoms the
b e tter for the system response time, but it may be more practical to install the
sample probe on the reflux accumulator, inlet feed or piping to product storage if
the desired stream is an overhead stream. As noted, the phase at the sample
location and the sample transport phase to the SCS must be carefully considered.

Obviously, the greater the difference in pressure from the sample point to the sample
return point the greater the sample flow rate and the faster the sample system can
deliver a representative sample to the GC in the vapor phase. From an analytical
point of view, sample vents should vent to atmosphere or atmospheric pressure.
The initial sample pressure for the depropanizer and debutanizer overheads is so
low that they may have to be returned to a flare stack. Any samples returning to a
low pressure flare should have check valves to eliminate reverse flow conditions into
the SCS when the plant goes to flare.

7
Vapor-phase samples can be transported more rapidly than liquid-phase samples
for decreased lag time. Sample velocity, through sample tubing, increases as the
volume decreases. An additional benefit is the reduced sample volume removed
from the process and returned to the low pressure sample return point. In some
cases the sample maytend to be two-phase. It is easier to heat two-phase samples
for transport in the vapor phase than it would be to cool them, or boost the pressure,
for transport in the liquid phase. Vapor phase sample transport is the preferred
method on the inlet stream and all the overhead streams (dethanizer, depropanizer
and debutanizer). The VPR shown at the sample point on the inlet stream is required
d u e to the Joules Thompson effect but heat tracing alone is adequate on the
remaining streams.

8
Care must be taken to ensure that sample transport tubing is heat-traced to about
70
C (160
F) and insulated depending on ambient temperature. Heat tracing must
be installed right up up to the entrance to the sample system oven to eliminate cold
spots on the tubing that would cool the vapor-phase sample and liquify the heavier
components.

Figure 2

9
Stream Switching

Once the sample has been filtered, pressure regulated to about 2-20 psig and
heated to maintain a single phase sample, stream switching can occur. Three
common design options for switching streams are:

1- Solenoid purge,
2- Double block and bleed, and
3- Single block.

Solenoid Purge: Solenoid purge designs are used to ensure that cross-stream
contamination does not occur in multi-stream sample systems. Off streams are
purged with a split flow from the selected stream.
The selected stream
e n e rg i ze s i ts
solenoid to allow
sample flow from the
normally closed
(N /C) port to the
common (C) port. A
split in the flow
occurs allowing the
sample to flow in two
directions. The first
path flows upward to
the sample valve for
analysis, the second
path flows from C to
normally open (N/O)
for the off streams to
carry potential leaks
from off stream
solenoids away from
the selected stream
flowing to the
sample valve. The
solenoid purge vent
carries the purge
flow from the sample
s ystem to a low
Figure 3 pressure return point
i n th e p rocess.
Each stream must
be vapor- phase and pressure balanced up to the stream select solenoids

10
The calibration standard uses a single block to conserve valuable calibration gas.
In addition, it is assumed that calibration gas is similar in composition to the selected
stream and potential leaks will not seriously bias the results.

Double Block and Bleed: A double block and bleed design also ensures that cross
stream contamination does not occur in multi-stream sample systems. The primary
source of cross-stream contamination is leaky solenoids and areas of unpurged
tubing common to
more than one stream.
This design allows
sample flow from the
selected stream to
pass through two
s o lenoids en route to
the sample valve. The
other streams are
double blocked with
two solenoids and a
l o w pressure vent to
a l low flow from either
p o te n ti a l l y leaky
solenoid to escape out
the bleed vent. The
selected stream
e n e r g i ze s b o th
solenoids to allow
sample flow from the
normally closed port
(N/C) to the common
port (C). The other
streams flow from N/O
to C. This method
uses more solenoids
Figure 4 than the solenoid purge
design to accomplish
the same result.

11
Single Block: The single block design does not take into account potential leaks from
s o l e noids. There is no stream-to-stream purging to eliminate cross-stream
contamination should one of
th e solenoids leak. This
design minimizes cost, but
should onlybe used in cases
where alternate streams are
very similar in composition
and SCS oven temperatures
are relatively low. Solenoids
that operate at higher
temperatures are more
prone to leaks and generally
experience a higher failure
rate. This design does not
purge samples other than
the sample flowing to the
sample valve. For example:
C a libration gas in figures
2 ,3 , and 4 is switched on
with a single block design to
minimize unnecessary loss
th rough a purge or bleed
vent. Small leaks across this
s o lenoid will mix similar
s a mples with a minimal
Figure 5 effect on readings.

Liquid Phase Samples

Sample Transport System for Liquids

To determine sample pressure for transport of a liquid-phase sample from the

probe to the GC, the full scale range of the lightest components must be used. This
ensures a representative sample will be delivered to the GC without “bubbling ” of the
lighter components under all process and environmental conditions.

For liquid streams there are basically three choices for the point at which the sample
is vaporized: the sample point, the sample conditioning system or the sample valve.
In figure 2, the inlet stream was vaporized at the sample point with a VPR and heat
tracing to the sample conditioning system. This greatly reduced the sample transport
time (lag time) and also reduced the amount of sample returned to the process.

12
Although this stream was liquid phase, in the process it was treated as a vapor
phase sample within the analytical system because the compositions were light
enough to allow for vaporization without excessive heat requirements. The remaining
column overheads and dethanizer bottoms streams in Figure 2, were sampled as
a vapor from the process and maintained in the vapor phase with heat tracing to the
SCS.

The heavier streams, (i.e. the depropanizer and debutanizer bottoms), are too heavy
to vaporize easily: temperatures in excess of 100
C would be required to maintain
a vapor-phase sample for the higher concentrations of these heavier components.
These streams are sampled from the process, transported to the sample system and
delivered to the sample valve in the liquid phase.

Liquid Phase Sample Inject

Depropanizer bottoms and debutanizer bottoms are sampled and transported in the
liquid phase at fairly high pressures and temperatures. As a result, care must be
taken that the lightest component, propane, does not bubble into the vapor phase
as a result of the higher temperatures. Any reductions in sample pressure may
cause propane to vaporize. In this case both bottoms products will be liquid phase
right up to the sample valve in the GC oven without pressure regulation until after the
sample valve. This back pressure regulator maintains a constant pressure in the
sample valve at the highest possible pressure to ensure a single phase liquid sample
for injection.

Any bubbling of lighter components at this point will affect the measured volume
i n j e cte d into the columns for measurement. This injected volume must be
repeatable from one analysis to another to ensure reliable, accurate results. In most
cases the process pressure is adequate for these types of liquid streams. If the
potential for bubbling is too great, the stream may be better suited to vapor-phase
sample handling.

Alow pressure sample return point will be required from the process with a large
enough differential to create flow in the sample transport system (figure 6). These
liquid samples remain in the liquid phase for sample injection. Vaporization occurs
a fte r sample injection when oven heat and a light carrier gas (usually helium )
combine to vaporize these heavy samples.

Verysignificant differences exist in design considerations for liquid sample systems

as opposed to gases. The primary difference centers around the fact that liquids are
less active than gases and do not adequately purge common tubing and a common
sample valve. For this reason each liquid stream should have its own isolated flow
path and sample valve. Gas molecules are much more active.

13
As a consequence, they are more suited to multi-stream sample systems where
stream-to-stream purging techniques like solenoid purge or double block and bleed
can adequately purge common tubing.

In figure 6 the manual block valves (BV-1 and BV-2) allow for manual introduction
of a common liquid calibration standard for each sample valve, as opposed to the
solenoid actuated calibration standard that was shown in figure 2. Manual control
is preferable on liquids because of an increased need for operator attention during
liquid calibrations to ensure accurate results. The back pressure regulators (BPR-1
and BPR-2) are required to maintain a suitable back pressure on the liquid sample
volume in the sample valve to ensure that a single phase liquid sample is injected.
As stated earlier, bubbling of the lighter components must not occur or the injected
volume will not be repeatable. Stream bypasses must return to a lower pressure to
ensure sample flow.

Sample flow and back pressure settings can be critical factors in determining the
sample transport lag time. Consideration must be given to liquid volume in the
sample transport system and distance the sample must travel to the sample valve.
For example, a high pressure liquid sample (400 psig) traveling 300 feet with a
stream bypass returning to the process at 370 psig is going to create a great deal of
l a g ti m e . As a result the overall system will have a very slow response time
regardless of how fast the GC is.

14
CALIBRATION STANDARDS

Vapor-Phase Calibration Standards

If the sample system produces two-phase samples en route to the GC, an error will
be introduced into the GC analysis that is dependent upon ambient climate
conditions. The same considerations must be given to vapor-phase calibration
standards and the sample transport tubing that carries the calibration standard to the
GC. These standards are mixed with the highest possible pressure based on the
lowest amb ient temperature stated. As a result, a heated calibration standard not
onlyensures that the heavier components are maintained in the vapor phase, but
contains more volume for virtually the same price. These calibration standards can
be verycostly, and care must be taken to ensure they are properly installed as cold
temperatures will liquify the heavier components and render them useless.

Figure 6

15
In colder regions a heated insulated cabinet is frequently attached on the outside of
an analyzer building. In warmer regions the calibration standards can be placed
outdoors provided the transport tubing and standards are heated to maintain the
temperature above the lowest ambient temperature specified for that mixture. To
co n s erve the calibration standard, 1/8 th inch sample transport lines are
recommended to reduce the volume required to purge the sample system with
sample prior to a calibration.

Liquid-Phase Calibration Standards

If the sample conditioning system supplies a liquid-phase sample to the sample valve
for injection in the liquid phase then the calibration standard must also be liquid
phase. The same is true for vapor-phase samples. Certified liquid calibration
standards, however, are considerably more expensive than certified gas standards.
In some cases, the user may choose to certify their own liquid standards form
laboratoryanalysis of liquid samples. Although this procedure may save on cost, any
errors in the collection of the samples or the subsequent laboratory analysis will
introduce errors into the process GC.

SUMMARY
To ensure reliable, accurate, timely results are available for process control the SCS
must be designed specifically for the process conditions on each individual stream.
An integrated approach to designing the sample transport system on site, to match
the sample conditioning system that arrives with the GC, will help ensure reliability
and faster analytical results. When GC results are in question calibration standards
can be used as a quick reference check for operations to easily verify on-line GC
results. Only then can the full benefits of on-line chromatographs be fully realized
to optimize plant performance and meet tighter product specifications.

Acknowledgments
Leighton Fields, Daniel Application Engineering . For providing engineering drawings & critical
review.
References
“Fractionation and Absorption,” Gas Processors Association, Engineering Data Book, Volume II,
Chapter 19.

16