Mercury Monitoring in the Oil and Gas Processing Industry Dr Warren T Corns R&D Manager P S Analytical www.psanalytical.com wtc@psanalytical.com

Bio and Company Background P S Analytical specialises in the ultra-low level detection of mercury, arsenic, selenium, antimony, tellurium and bismuth in all sample types. Founded by Prof Peter B Stockwell in 1983. Dr Warren Corns joined in 1988 as PSA PhD student and became R&D Manager in 1991. Focus on R&D sponsoring eight PhD students and several postdocs over last 25 years Main Office is located in Orpington, Kent, UK . 40 staff involved in production, design, service, R&D, sales, technical support, consultancy and administration. US Offices are located in Deerfield Beech and New Jersey. SE Asia office in Queensland. Other countries supported by distributers Main focus of our business is Hg analysis with laboratory and online measurement equipment. The oil, gas and petrochemical industry is just one of our market areas. Huge focus on environmental applications.

Oil, Gas & Water Atmospheric Emissions Gas Oil/Condensate Solid Waste Flaring, vents & exhausts Gas 0.05 – 5000 µg Nm-3 Oil/Condensate 0.1 – 20 000 µg kg-1 Solid Waste e.g. drilling muds /sludge Water 0.1 - 50 µg L-1 Gas Oil Water Naturally Occurring Hg in all phases

A typical Gas Processing Plant

Amalgam Corrosion Amalgam corrosion occurs when Hg and Al create amalgam in humidity. Hg + Al  Hg(Al) (amalgamation)…………………….1 Hg(Al) + 6 H2O  Al2O3.3H2O + 3 H2 + Hg…………..2 Hg + Al  Hg(Al) (amalgamation)…………………...1 Due to Hg regenerating itself, the reaction is self propagating while water exist . The amalgam corrosion will accelerate in presence of oxygen, creating a corrosion of Al2O3 feathers

Liquid Metal Embrittlement LME is the most important of the failure mechanisms that attack aluminium in gas plants. Amalgamation occurs in a grain boundary followed by an inter-granular stress cracking activated by stress and or residual stress. No water is required, however if water is present the corrosion is accelerated. In case of aluminium alloys this attack has been observed where magnesium precipitates in grain boundary during equipment manufacturing process or welding process. 4Hg + Al3Mg2 = 2MgHg2 + 3Al For effective LME attack, the Hg needs to be in liquid phase above the freezing point of Hg (-38.9C) and also exist as a crack in the aluminium protective oxidized film.

Liquid Metal Embrittlement

LNG Plant in Algiers. Explosion by mercury induced heat exchanger failure.

Santos LNG Moomba Plant 1st Jan 2004, Santos LNG plant in Moomba, Australia explode. Cause of loss, failure of a nozzle on a heat exchanger from Liquid Metal Embrittlement Fatalities : Resulted in the interruption to gas supplies in South Australia and the Eastern Seaboard No Injuries Damage to Plant substantial and loss of revenue 7 months of gas interruption supply On the 23 December 2005 the claim was settled for A$231 Million (From : “What happens when the catastrophe occurs” Chris McMichael, Presentation on behalf of Santos Ltd)

Official Methods for Mercury in Natural Gas ISO 6978, part 1 and 2. Part 2 is based on two heated gold traps in series to collect Hg. Measurement is based on dual amalgamation – AAS or AFS. ASTM 6350. Based on two gold traps in series to collect Hg. Measurement is based on dual amalgamation – AFS. PSA online and offline equipment complies to both methods. Detection limits are less than 1ng/m3.

Pressure Let down System for Offline Sampling

Stainless Steel IP66 Enclosure for 10.547 – Version for heated lines

Hg in Gas Analyser - Sir Galahad II

TGM Manual Measurements of Remote Tubes

Optical Configuration for AFS Plan View Mercury Lamp Lens Collimator + Filter Introduction Chimney Sample in Sheath Argon Reference cell Photo- multiplier tube

Vapour Injection Calibration Procedure Capillary Tube Thermometer Gas Tight Syringe Elemental Hg B C A

Typical results using Hg removal beds Outlet Specification < 10ng/Nm3 Sample Description Bed Volume m3 Inlet Conc. ng/Nm3 Outlet Conc. Sales Gas 31 279 ± 12 370 ± 53 735 ± 130 522 ± 72 3.2 ±0.9 Dec 11 2.6 ±1.1 Aug 11 7.1±0.9 Dec 10 5.9 ± 2.6 July 10 10 1268 ±197 1.2±0.7 Regeneration Gas 3 370 ± 139 1.1 ± 0.1

Amasil Trap Cleaner

How do you know if your results are good? Sampling system is well designed so that no adsorption losses or contamination is experienced. Check blank on sampler using purified air/nitrogen. Check reference gas through sampling arrangement Sampling system and sample point is flushed with sample prior to use Bypasses are used to ensure a representative sample is obtained Sample traps are regularly checked by spiking known masses of Hg before use Ensure calibrated and certified components are used throughout Ensure that no condensation is experienced during sampling. Use phase diagram. Traps must be used in series so that breakthrough can be monitored (<10%) Traps are heated (circa 100oC)to ensure no condensation on gold Test different sample volumes to ensure result is independent on volume collected Dynamic spike additions in the field to check collection and efficiency in presence of sample Participate in inter-laboratory collaborative and proficiency exercises

How about Wet Gas? More challenging for a number of reasons Gas saturated with water at process conditions High potential of liquid hydrocarbon (BTEX) and water condensation because of JT cooling at sampling arrangement High concentration of Hg – small sample volumes difficult to collect in practice Adjustment of primary bypass flow control can be problematic.

Arrangement for Wet Gas Deliver sample at line pressure via a short Teflon braided hose Control primary bypass flows using calibrated armoured flowmeters at process conditions High flow bypass upstream of primary pressure reduction. Condensation is only in vent line so does not affect measurement Primary pressure reduction with a heated regulator to overcome JT cooling prevent condensation – Dropping the pressure in stages is standard practice to avoid condensation Entire sampling system is coated with SilcoNert to avoid losses of Hg and minimize carryover

Sampling arrangement for wet raw gas

Experiences with Wet Gas Sampling arrangement All lines in sample pathway run warm well above the dew-point of sample even with multiple sampling systems connected. Results for wet gas look very good showing stable results with good repeatability. Short term precision typically below 5%RSD No breakthrough on traps Results were independent of volume Gas Analysis Location Time Volume Collected (L) Result (ug/Nm3) % Breakthrough 22/10/2016 Train 1 Outlet 13:30 - 13:45 10 149.1 1.2 16:35 - 16:39 2 148.3 0.8 17:10 - 17:16 3 151.7 0.9

20.630 LPG Accessory

Manual Sampling onto Remote Traps

European Mercury Removal Plant Yearly Average in RED

Mixed LPG Plant Case Study Reported Hg Issues for C4

Mercury Contribution from Offshore facility in Indonesia Site produces at maximum capacity 1 Billion Cubic Feet of Mixed LPG/day equivalent to 28,000,000,000 litres/day At Average Concentration of 1.2mg Hg/m3 this equates to a Hg loading of 33.6Kg Hg/day (12.3 tonnes Hg per year) Low Hg in LPG due to storage at low temperature (freezing point Hg = - 39oC) Stabilized condensate 38 ± 3 ppbw

Hg Condensation Model

Cold Box and Chiller trains

Hg in Produced Water It is estimated that more than 50% of the Hg released from the oil and gas industry is from produced water. Environment and Regulation bodies enforce different permitted discharge concentrations mostly between 0.1 to 10 µg/L. Not strictly enforced in all areas. To achieve this concentration some sites require Hg removal.

Millennium Merlin – Cold Vapour AFS System Elemental, inorganic and organic mercury forms present in sample . Chemical digestion with acidic oxidants required before CV-AFS. Stannous chloride reduction of Hg2+ to Hgo Detection Limits less than 1 part per trillion. Linearity 5 orders of magnitude 2 minute measurement time

Typical Results for Hg in Seawater and Produced Water Sample Reference Concentration (ug/l) % Spike Recovery Water CRM (ORMS-1) 0.0066 ± 0.0003 (0.0068 ±0.0013) 96 - 102 Formation Water – North Sea 14.7 ± 0.5 95 - 98 Produced Water - Canada 0.57 ± 0.02 96 11.66 ± 0.01 98 Produced Water – SE Asia 2098 ± 136 92-103 Produced Water – North Sea 3.24 ± 0.01 94

Hg in Crude Oil and Condensates Very wide Hg concentration range from low ppb to ppm Raw crude and condensates requires processing. Desalting and caustic wash treatment both remove Hg During stabilisation of raw crude, elemental mercury is released to overheads. During processing Hg has a tendency to pre-concentrate in lighter fractions producing Hg concentrations in excess of raw feedstock. Hg becomes distributed to all refined products. Mass balances are difficult because of plant losses.

Relative Weight Distribution of Hg in Condensate Cuts – Data from IFP

Mercury Removal for Liquid Hydrocarbons Liquid feed-stocks (propane, butane, naphtha, condensates) used by the petrochemical industry are typically treated to remove Hg to below 1ppbw. Other heavier streams such as diesel, kerosene, fuel oil and residuals are not treated or tested for Hg. Typically used as fuels. Mercury is known to cause problems in the petrochemical industry due to poisoning of expensive precious metal hydrogenation catalysts and Hg attack on heat exchangers Removal technologies are similar to gas applications (e.g sulphided metal oxides).

Commonly used Measurement Techniques for Hg in Liquid Hydrocarbons Chemical Digestion/Extraction with CVAFS Volatilization – Amalgamation -AFS (Suitable for Online Measurements)

Typical Results for HgTotal by Aqua Regia Sample Prep Mercury (ng/ml) Mean Result (ng/ml) Conostan CRM 1 2 35.3 ± 0.3 33.7 ± 0.3 34.5 ± 0.3 (100%)* North Sea Crude Oil 1 97.1 ± 0.2 103.6 ± 0.5 100.4 ± 0.5 (97.2%)* North Sea Crude Oil 2 78.5 ± 0.1 74.3 ± 0.3 76.4 ± 0.3 (102.1%)* Thai Crude Oil 2896 ± 15 2991 ± 8 2944 ± 17 (96.4)* Venezuelan Crude Oil 3852 ± 18 3926 ± 10 3889 ± 21 (95.1)* Algerian Condensate 733 ± 14 741 ± 15 737 ± 21 (98.2)* * Spike Recovery Data, MDL=0.2ng/ml

Schematic for Automated Mercury Measurement for Liquid Hydrocarbons

Comparison of Data between Digestion CV-AFS and Volatilization - AFS Sample Reference Hg ppbw using 10.515 Hg ppbw using CV-AFS NWS Condensate 36.8 +/- 1.3 35.1 +/- 0.5 Saudi Aramco Condensate 15.7 +/- 0.7 16.9 +/- 1.0 Malaysian Condensate 56.9 +/- 2.8 59.4 +/- 0.1 Condensate Outlet MRU 0.14 +/- 0.03 >0.2 Naphtha (UK) 5.65 +/- 0.12 6.21 +/- 0.23 Naphtha (Hawaii) 265 +/- 12 260 +/- 2.5 Norwegian Condensate 2.51 +/- 0.07 2.20 +/- 0.13 Australian Condensate 36.0 + 3.7 35.6 +/- 2.4 (pyrolysis =38) South African Condensate 1.81+/- 0.12 2.01 +/- 0.25

Hg Species in Liquid Hydrocarbons? Mercury Classification Forms of Mercury Elemental Hg Hgo(gas), Hgo (dissolved) , Hgo (solids) Particulate Hg HgS, Hg (solid associated), Hg (insoluble) Inorganic Hg Hg2+, Hg22+ Organometallic Hg R-Hg+ ,R-Hg-R, Hg thiolates - Unknowns?? Complexity of sample matrix and number of potential species of Hg inhibits the use of a single approach for determining all Hg species Fractionation Methods are therefore used

The Fractionation Method Hg Total CV-AFS Particulate Hg HgS Aqua Regia Other Particulate Hg Nitric Acid Volatile Hg Purge & Trap Dissolved Hg KCl extraction Ionic Hg Aqueous Phase Unknown Hg Organic Phase Organic Hg GC-AFS

Hg Fractionation in Condensates Example data Sample A Sample B Sample C Sample D Sample E Sample F Sample G Sample H Sample I Total Hg 1.00 1.24 780.80 8.17 420.74 73.73 168.30 1034.00 925.30 Dissolved Hg 0.73 0.66 552.70 0.00 19.82 0.44 0.50 14.00 2.30 Ionic Hg, KCl 0.57 0.36 13.60 3.95 18.59 0.18 0.23 0.25 Unknown Hg, non - KCl NA 539.10 Organic Hg Particulate Hg 0.27 0.58 226.20 4.06 400.92 73.29 167.80 1020.00 923.00 Volatile Hg 0.10 0.07 1.90 0.01 0.06 0.84 0.05 0.24 0.35 Fraction Sum 0.94 (94%) (100%) 8.02 (98%) 419.57 74.31 (101%) 168.03 1020.47 (99%) 923.60 NA = Not Analysed - Only analysed if summation of fractions is poor ND = Not detected, MDL= 1ng/g All results in ng/g (ppbw) Fraction Sum = summation of columns in yellow Results have higher confidence if total Hg and Fraction Sum are in good agreement

Hg Fractionation in Crude Oil Example Data B1 B2 B3 B4 B5 B6 B7 B8 Total Hg 538.24 98.79 60.80 37.04 62.56 46.79 42.38 84.83 96.00 Dissolved Hg 43.44 7.72 9.12 4.25 11.82 7.79 6.69 13.13 4.99 Ionic Hg, KCl 19.06 4.73 7.59 3.43 6.08 4.46 3.82 6.60 0.74 Unknown Hg non-KCl NA Organic Hg ND Particulate Hg 494.80 91.07 51.68 32.79 50.74 39.00 35.69 71.70 91.01 Volatile Hg 24.25 0.20 0.10 0.08 0.12 0.33 0.25 0.37 Fraction Sum 538.11 (100%) (97%) 59.37 (98%) 36.30 57.02 (91%) 39.12 (84%) 39.84 (94%) 78.55 (93%) 92.12 (96%) NA = Not Analysed - Only analysed if summation of fractions is poor ND = Not detected. MDL = 1ng/g All results in ng/g (ppbw) Fraction Sum = summation of columns in yellow Results have higher confidence if total Hg and Fraction Sum are in good agreement Sample B – Same plant sampled at different periods

Sampling Uncertainties for Liquid Hydrocarbons Depressurization and stabilization might release volatile Hg compounds such as elemental mercury vapour. Potential Loss of Mercury on sampling components. Materials and speed loop should be carefully considered. Losses of Mercury from association with particulates and aqueous phase. Sample filtration devices and sample point selection should be carefully considered. Losses and species conversion of Mercury during storage and shipping of samples. Sample containers and storage conditions are critical. Sample homogenization is required prior to measurement. Shipping regulations often require metal containers!

PSA 20.610 Auto injection of liquid hydrocarbons to 10.515

What type of Samples? High pressure samples collected in Bombs 20.610 – 10barg 20.610H – 100barg Samples that are compatible with the 10.515 Hg pre-concentration unit Samples that are 100% liquid at process conditions. This information is obtained using a Phase Diagram. Un-stabilized condensates, un-stabilized crude, depropanizer bottoms and debutanizer bottoms, LPG, propane, NGL, butane, naphtha, etc. Collecting samples in open vessels will allow gases to vent off and causes losses of elemental Hg. There may be cases where higher temperature cause gasification in the sample bomb

Phase Diagram @1bar, 5°C, sample is 50.6% gas. @20°C, >3.5 bar g needed to keep sample in liquid phase.

20.610 Auto injection system

Mass Balance Ethylene Plant PS Analytical are an instrument manufacturer specializing in the measurements of Hg and Hydride forming elements using AFS. We also provide consultancy onsite and offsite measurement services. PSA were approached by a Olefin plant to independently test for Hg across the plant because they were experiencing poisoning of Palladium based acetylene reactor catalysts. The spent catalyst was tested for Hg (40-60ppmw) which confirmed Hg was the issue. The site added a UOP Hg removal control to the Mol Sieve gas and liquid dryers and replaced the acetylene catalyst. PSA visited site to perform Hg testing of the feedstock streams and also to test the performance of the newly installed mercury removal units. Other streams were tested to gain better understanding of the fate of Hg across the plant.

Olefin Production

Sample Types Gases and Liquified Gases (e.g. C1-C4) – Sir Galahad Liquid Hydrocarbons (e.g. C5+) – 10.515 Hg preconcentrator Aqueous Phase (e.g. caustic wash/wastewater)- CVAFS Gas – Liquid Mixtures (e.g. DeC3 bottoms) – Combined

Multiphase Samples Gas Analysis Liquid Analysis

Mass Balance Model All samples measured and collected over a 7 day period at steady state conditions. 3 gas samples strategically positioned so that each section of the plant could be studied. Liquid samples collected every day am and pm. Main source of mercury to the plant was from the butane feed. Gas phase measurements generated results in ng/Nm3 so results must be converted to ppbw (µg/Kg = mg Hg/Tonne). Process flows recorded in Tonne/Hour. Mean flow over sampling times was used for calculations Mass Balance calculations were based on mg Hg/Hour All measurements done on site on samples collected within 24 hours if possible.

Results

Mercury Removal Unit

Mercury Removal Results Gas Dryer MRU efficiency = 99.7% Liquid Dryer MRU efficiency = 94.2%

Acetylene Reactor Beds Acetylene Reactor Inlet should ideally be less than 10ng/m3 Mercury Removal Units only recently installed so quite possible that Hg result is elevated from pipework contamination Testing several months later gave results less than 10ng/m3

Mass Balance Summary A mercury mass balance was successfully performed on Olefin plant producing ethylene and propylene using atomic fluorescence spectrometry for gas and liquid phase samples Hg was found in the butane feedstock at 18.3ppbw (46350ng/m3) 15% of this Hg was removed at gasoline stripper bottoms 4% and spent caustic 11% The remaining 85% of this Hg was split between gas (93%) and liquids (7%) A mass balance close to 100% was found between feed and gas/liquid dryers. Mercury removal beds installed at the gas and liquid dryers removed 99.7 and 94% respectively Downstream acetylene reactor and cold boxes considered low risk although not quite operating at ideal specification of less than 10ng/m3 A model estimating distribution of Hg downstream was produced which suggested low levels of mercury across fractionation trains and splitters

Online Mercury Analyser

Block Diagram of Online Natural Gas System

Gas Stream Sampling

Hg Measurement Amalgamation AFS

Online Determination of Mercury in LNG

Online Liquid Stream Analysis

Liquid Stream Sampling

Online Hg in Naphtha Results from MRU with Zero and 10ppbw Conostan Span Checks

Combined Online Gas and Liquids

Performance Specifications

Development of Hg Removal Adsorbents

Development of Hg Removal Adsorbents

Development of Adsorbents for Liquid Hydrocarbons Cycling feed of liquid hydrocarbon dosed with elemental Hg Metering pumps to log flow across beds 4 test beds Direct measurement of inlet 1ppm and outlet streams (<1ppb) Auto-calibration

Occupational Hg Exposures The oil, gas and petrochemical industry have a various Hg occupational exposures to consider. Condensation of Liquid Mercury. This can equate to several tons of Hg being drained from the process. Hg contamination on pipework and vessels Hg in various exhausts and re-boiler vents Hg in spent adsorbents, catalysts Hg in waste products such as sludge, caustic wash, spent amines, spent glycol and wastewater Hg on PIG receivers Hg in various refined product streams and also vapours. Exposure Management by air monitoring, staff urine and blood tests, education HSE training.

CONCLUSIONS This presentation has hopefully provided an overview of monitoring techniques for Hg in the Oil and Gas Industry. Additional research needs to be done to study the behaviour of Hg during the extraction and processing oil and gas. Mercury removal approaches are very effective but they are focused on mercury removal on specific product streams to meet specifications and to protect downstream components. Offline and online Hg measurement technologies are available from PS Analytical for all types of samples for the oil, gas and petrochemical industry.

Thank you for attention. Any Questions?