isa 2008 remote injection montgomery
TRANSCRIPT
Analytical Solutions for Process Control & ComplianceThe 53rd Annual Symposium of the Analysis DivisionCalgary, Alberta Canada; 20-24 April 2008
PROCESS GC ANALYSIS OF A HAZARDOUS SAMPLE USING A
DISCRETE SAMPLE VALVE IN A REMOTE LOCATION
Tom G. MontgomerySiemens Energy & Automation, Inc.
Process Analytics
Presented in Calgary by Bob Farmer, SE&A
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 2
Outline
• Introduction and Abstract
• Background and the project requirement
• Solution to project requirements
• Possible issues with solution and investigations
• Result
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 3
Introduction and Abstract
• Practical example of discrete remote sample injection– Many thanks to Dr. Jimmy Converse
• Owner’s requirement– Hydrocarbon components in hazardous stream matrix– The owner made severe constraints regarding sample handling
• Solution– Description of remote discrete sample injection technique– Discussion of potential negative impacts
• Investigations– Discussion of possible negative effects on the measurement
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 4
Background
• GC is a sample extractive technique– To minimize lag time, a “large” quantity
of sample is removed from process and transported to the analyzer at high flow rates
– At the analyzer, a small quantity of sample is extracted again and conditioned for presentation to the analyzer at low flow rates
– Unused sample is returned to process or to flare
– In the analyzer a tiny quantity of sample is extracted again and inserted onto the columns
• Original extraction and transport issues
– Flow 1: typical 3.5 l/min to ½” tubing– Flow 2: typical 30-40 ml/min to 1/16” or
1/8” tubing– Sample 3: typical 0.5 µl sample or less
Process Stream
ElectronicsOven
Carrier Gas
Gas Chromatograph
Sampling System
Sample Valve
Columns and Valves
Detector
Temperature Controlled
2
1
3
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 5
Standard GC Extractive Sampling Technique
Illustration Ref. Dr. J. Converse
Analyzer ShelterUnused sample to flare or return to process
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 6
Benefits of Standard Technique
• Ensures fresh sample is present at inlet of GC (reduces time lag)– In this application, transport-related lag time = ~30 sec
• No affect on chromatographic method
Analyzer ShelterUnused sample to flare or return to process
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 7
Problems of Standard Technique
• Extracts large quantities of sample from process
• Excess sample must be disposed or returned to process
• Failure conditions, including leakage on the transport line must be accommodated
• Leakage can result in hazards of fire or injury to personnel– On this project, the stream contained ~2% Hydrofluoric Acid (HF)– Lethal to humans: STEL <3ppm; Extremely corrosive
Analyzer ShelterUnused sample to flare or return to process
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 8
Solution – Remote Discrete Sampling
Illustration Ref. Dr. J. Converse
Analyzer Shelter
Sample valve located near process
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 9
Benefits of Remote Sampling
• Eliminates transport of large quantities of sample
• Minimizes waste
• Reduces issues of leakage away from process– Reduces human hazards– Reduces potential for damage
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 10
Potential Negative Effects on Analysis
• Increased cycle time (or lag time)
• Retention time shift– Effect on valve operating times– Effect on gating reliability
• Peak effects– Peak broadening– Measurement performance
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 11
Transport Tubing “Part Of Oven”
45
21
36
45
21
36
Vents
Car. Ref.
Car. 2
Car. 1
Sample fromProcess
KOh
Ref BF Main ITC
Oven Temp: 60°C ±0.05°CValves and wetted tubing of Hastelloy C
C1b
C2
C3
60 Ft
Transport Tubing must be at a controlled temperature: ±?°
“The heated transfer line can be perceived as a very long, slender oven.”(Dr. J. Converse)
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 12
Questions To Be Investigated
• How much lag time is introduced?
• How good does the temperature control of transport tubing have to be?– Effect on Backflush Valve operation– Effect on gating of components
• Any other analytical issues?
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 13
Test Set
• Maxum GC with divided oven– Two isothermal halves– Each temperature
independently controlled
• Application with SV and CV and columns at 60°C (analysis temperature)
• Coil (60’) of 0.032” ID / 1/16” OD tubing to simulate process transport tubing on other side
• Carrier pressure 51 psig and resulting flow 30 ml/min
• Varied coil temperature from 40° to 80°C in 10°C increments and plotted chromatograms
2 valves
Detector
Divided Oven
Application Side
Test Coil Side
(Photos representative of actual test set.)
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 14
Questions To Be Investigated
• How much lag time is introduced?
• How good does the temperature control of transport tubing have to be?– Effect on Backflush Valve operation– Effect on gating of components
• Any other analytical issues?
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 15
Lag Time Observed
Ref. Propane:Elution: ~42sec 151secAdditional Lag: ~109sec
ITC Without Tubing
ITC With Tubing at 60°C
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 16
Questions To Be Investigated
• How much lag time is introduced?
• How good does the temperature control of transport tubing have to be?– Effect on Backflush Valve operation– Effect on gating of components
• Any other analytical issues?
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 17
How Much Shift in BF Time Can Be Allowed?Same as previous slide with BF time set out to 240 sec.
Ref. Propane elution = 151sec.
Acceptable retention time shift: <±3 sec.
ITC With Tubing at 60°C
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 18
How Much Retention Shift Is Acceptable for Gating?
• Adequate separation suggests retention time shifts of ~10 seconds can be accommodated by normal gating technique.
• Therefore, the BF time requirement will be the primary control. (<±3 sec)
Main Detector – No Tubing
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 19
A Quick Test At the ITCITC With Tubing at 60°C
ITC With Tubing at 80°C
Retention Times:40°C : 177 (not shown)60°C : 171.380°C : 166.3
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 20
Main Detector; Tubing 60°C
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 21
Main Detector; Tubing 80°C
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 22
Main Detector; Tubing 40°C
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 23
Retention Time Effects Summary
• ±20°C shift in temperature of transport tubing causes retention time changes of >10 sec
• Transport tubing temperature should be controlled to at least ±5°C to ensure reliable performance of the analyzer
• (Note: Sample Valve temperature may also vary ±5°C since this is a liquid sample.)
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 24
Questions To Be Investigated
• How much lag time is introduced?
• How good does the temperature control of transport tubing have to be?– Effect on Backflush Valve operation– Effect on gating of components
• Any other analytical issues?
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 25
Peak Shape
• Columns used: 0.53 mm ID “megabore” capillary– Normal response is sharp (Eg; Methane ~5 sec wide
even when retained 2.5 minutes at 40ml/min flow rate)
– Therefore, negligible peak broadening was expected
• No significant broadening was observed
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 26
Measurement Performance
• Due to good peak shape performance, measurement performance is expected to be good
• All peaks performed well except C5+ (backflush peak) 40C 7.46%
50C 7.57%
60C 7.69%
70C 7.944%
80C 8.086%
10°C swing in transport tubing temperature cause about a relative 1.6% change in the C5+ concentration.
Note: C5+ in this sample is made of Isopentane through n-Hexane. May be a result of backflush of C5s and possibly C4s at lower temperatures.
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 27
Summary Of Observations
• Lag Time– Consistent with predictions from flow calculations– Acceptable in overall system performance
• Retention Shift– Tubing must be kept at a constant temperature ±5°C
• Analytical Issues– Negligible in some cases – but poor performance of
backflush “plus” peak also requires tubing to be kept at a constant temperature ±5°C
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 28
Results
• Remote discrete sampling resolves problems of hazardous sample handling
• Lag time effects are predictable
• Transport tubing must be kept at a controlled temperature – but not as tightly controlled as an analytical oven
• Measurement performance effects are acceptable if transport tubing temperature is controlled
• Additional benefit: HF neutralization can be accomplished with a small column instead of a large pot
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 29
Other Limitations
• This test was with a liquid sample– Simpler transport of sample aliquot– Simpler temperature control requirements on sample
valve (and transport tubing)
• Distance from process to analyzer was not large (~60ft)– Lag time increase is acceptable even at relatively low
flow rates of chromatographic columns
AD 2008: Analysis Division SymposiumCalgary, AB Canada; 20-24 April 2008
Slide 30
Acknowledgements
• Dr. Jimmy Converse (research and publications)
• Steve Trimble (Siemens Process Analytics)