Naphtha Product Spill and Leak Detection in Shared Process Cooling Water
David J. Veltkamp - Presenter CPAC, University of WA John Crandall Analytical Specialists, Inc. Brian Rohrback Infometrix, Inc.
Define the Problem
• Hydrocarbon spills & leaks are costly events resulting in – Loss of product to waste – Manpower and other resources to correct the problem – Damage to the environment – Possibly regulatory and public relations issues – All leading to lost productivity and profits Prevention of “spill & leak events” – Is critical to process operation whether – Into the air, process or waste water or on the ground Inevitably, spills and leaks will occur requiring – A system for detection and – Diversion of the material for correction and remediation
•
•
Focus of This Work, Naphtha Products in Water
• Naphtha range products including – C6 to C10 hydrocarbons – With varying volatility and solubility in water • Production shares cooling towers – Multiple naphtha product types use the same heat exchange system – Once through cooling water shares the same discharge point • Thus, simple leak detection using wastewater total organic carbon (TOC) is unsuitable – No speciation for product type spilled or leaking – Varying, temperature dependent solubility of the naphtha products lead to erroneous results
Measurement Possibilities
• Gas Chromatography with Flame Ionization Detection – Should provide the speciation required for leak identification – Direct liquid injection would simply duplicate TOC but with speciation – May not be sensitive or fast enough for the requirements • microGC and FID coupled with trapping technology – Provides more sensitivity with the speciation & speed required – Sensitivity enhancements enable headspace gas sampling • Headspace gas sampling removes “particulates” from the sampling problem and • Turbulent water effectively “sparges” hydrocarbons from the water enriching the headspace gas with naphtha sample minimizing temperature variability and erroneous measurements
Measurement Possibilities
• microGC FID with trapping coupled with chemometrics – Provides fast sensitive product leak detection and information for initiating diversion to the remediation tank – Provides faster, easier product leak identification – Can lead process operations to the source of the leak quickly for repairs • The possibilities needed validation – Can the microGC detect naphatha components? – Can the microGC chromatographic results identify the components? – Can the microGC and chemometrics discriminate “background within a hydrocarbon processing facility” from “leaks?”
On Site Measurement Validation Project Definition
• • Establish “normal background” Determine whether new microGC solution can work – Demonstrate that the background is sufficiently low – Demonstrate that the “detect & divert” measurement levels are sufficiently high enough to differentiate from background – Demonstrate “leaked product identification” potential – Demonstrate chemometric application efficacy • Onsite work definition – Measure background at “sump” and through a “sump sampling system” – Measure prepared naphtha product samples at the “detect and divert” levels • Follow up with data reduction and presentation
microFAST GCtm with EZChrom Used for On Site Validation
System Operation
Typical Results, Solvesso 200 ULN in Wastewater Headspace Gas
Simultaneous, 2 Channel FID results in 75 seconds Column 1, DB5
Column 2, DB1701
Typical Naphtha Product Range (in Wastewater) Results
Typical Naphtha Product Range (in Wastewater) Results
Discussion of Results
• Typical Naphtha Range products – Were prepared in water at the detect and divert levels – Solvesso 200 ULN was chosen for special attention • Has the highest water solubility translating into the lowest wastewater headspace gas concentration • Deemed to be the most difficult of the range to discern over normal background • Chromatographic traces are distinctive for each naphtha product – Boiling range distributions follow expected product compositions – Sensitivity over noise is more than adequate – But is it adequately higher than “background?”
Two Wastewater Sewer Background Chromatograms, original scale ~15,000 F.S.
Cyan sample from “sample system” Blue sample manually collected with a syringe
Cyan sample from “sample system”
Red sample manually collected with a syringe
Two Wastewater Sewer Background Chromatograms, at Solvesso 200 ULN scale
Solvesso 200 ULN at Detect & Divert Level Overlaid Wastewater Background
Blue is wastewater background. Red is Solvesso 200 ULN.
Solvesso 200 ULN at Detect & Divert Level Overlaid Wastewater Background
Cyan sample is Solvesso 200 ULN Blue sample is background
Cyan sample is Solvesso 200 ULN Red sample is background
Discussion of Results
• Solvesso 200 ULN – Has the lowest headspace gas concentration – Is the most difficult to measure – Yet is clearly distinctive and well above “normal background” at the detect and divert level – Plenty of sensitivity is available over background for early warning • All naphtha range samples – Were prepared at the detect and divert levels for each product – Headspace gas concentrations range from 0.4 to about 10 ppm – All are clearly distinctive, one from the others and – Well above background levels
Chemometric Modeling Approach
• • It is clear from the analyses that the four different naphtha products can be differentiated (visually) by GC We want to demonstrate that the chemometrics can reliably differentiate products – Lump all chromatograms into a single data file – Apply Principal Component Analysis (PCA) to identify and model chromatographic characteristics that distinguish different possible products that could potentially leak into waste water – New chromatograms projected into PCA model to identify • Has waste stream composition changed? • Can we identify what changed (products in the waste stream)? • Can we determine severity of leak?
Pattern Recognition (PCA Modeling)
2 Clusters of Wastewater Air Because these 2 are mislabeled blanks. Solvesso 200 ULN
PCA scores differentiate products
Aligned
The scores trend is explained by concentration differences
Unaligned
The scores trend is explained by concentration differences
Aligned
Discussion
• Chemometrics Results – Demonstrates the separation in the samples of concern and – That it is unlikely that samples will be mis-identied even when mislabeled – These techniques can be used not only for the analytical results but also as “system diagnostics” – For example, sample system failures (like a sticking sample valve) would produce chromatograms and PCA plot positions that look like “blanks.” – Flags could be set to notify personnel of a problem and what the likely problem and solution might be • With only a little more work, should be possible to implement PCAbased control charts and /or on-line monitors for contaminate levels
Conclusions
• Micro GC coupled with trapping technology – Is sufficiently sensitive for the job – Can detect “worst case” leak situation for the high solubility product and Micro GC coupled with chemometrics – Is sufficiently selective for the job – Can differentiate the hydrocarbon products made in the plant and use this information to identify the source of any leak – With more data, we can build a mixing model and quantitate the relative contributions of multiple sources – All interpretations and quantitations are automated and can be fed into any control system Micro GC coupled with NeSSI™ is a good fit – Eases implementation: feasibility ↔ deployment Feasibility is one thing; this system needs to go on-line