WMS Wellhead Metering Systems Sampling Training Course

Transcription

1 NOTES ON AUTOMATIC SAMPLING Bill Topham BSc CEng FIEE FinstMC - Technical Adviser Bob Maurer BSc MSc C.Eng. MinstMC - Wellhead Metering Systems This section may be reproduced in whole or part with acknowledgement to Bill Topham and Bob Maurer for training a new generation of Instrument Engineers who were not involved in the original development work CONTENTS Page 2 Page 3 Page 4 Page 5 Page 6 Page 7 Page 8 Page 9 Page 10 Page 11 Page 12 Design philosophy of WMS range of balanced seal samplers Reasons for sampling and its major applications Origins and development of automatic sampling Water profile in pipelines pipeline mixing types of mixers Taking representative samples design criteria for probes Principles of grab sampling time and flow proportional modes Types of grab samplers In-the-Line (ITL) and Fast Loop/Cell Types of sample receivers fixed volume and variable volume Comments on sampling systems Sample handling, mixing, sub-sampling and lab testing Continuous measurement of water content About the writers: Bill Topham worked for British Petroleum Company (BP) for 34 years. His work covered all aspects of measurement and control but specialised in developing on-line analysis equipment and techniques. The BP team in conjunction with R. Maurer developed the Crude Oil sampler which was originally licensed to Maurer. A separate development programme was carried out with Jiskoot Autocontrol. Bob Maurer separately developed for BP. Exp. the Piston Internal Mixing Receivers, PVT Receivers, Scoop Tube, CruFlo Flow meter and supplied the original Jet Mix Nozzle which formed the basis of the ISO 3171 Standard 1

2 Design parameters of WMS Crude oil Samplers to meet the requirements for accurate sampling of two phase fluids i.e. crude oil /water mixtures The representative sampling of crude oil containing water is clearly set out in ISO Standard 3171 and equivalent Codes of Practice. As the final amounts of collected sample used for determining the water and calorific content of an oil consignment is of the order of 0.3x10-6 % it is of vital importance that the sampling system is one in which the biasing of either the water or oil content cannot be tolerated. This specific requirement led to BP Exploration requesting Bob Maurer in 1982 to design and build the sampler that would grab a fully representative sample with ZERO BIAS using the BP patented 3 step sampling concept. 1 Isolate a representative cross section of the flow stream 2 Pump the trapped sample into an atmospheric receiver for stabilised crudes or into a pressurised receiver for live crudes 3 The Capture Tube and the Plunger Piston returns to the ready for the next sample grab The present WMS range of Crude Oil Samplers include advanced design variations without affecting their basic function and replacement compatibility of samplers presently in operation. The major difference lies in taking advantage of modern machining techniques to introduce ports that allow for balanced seal systems to be employed. In the case of the electric motor driven samplers this has allowed for the same small size 0.09KW watt motor / gearbox system to be employed for all pressure ratings from ANSI 150Lbs to ANSI 2500 Lbs. without loss of sample grab repetition rate. In a similar manner the low pressure (10 Barg) pneumatic actuator also remains unchanged throughout the entire pipeline pressure range of ANSI 150 Lbs to ANSI 2500 lbs 2

3 REASONS FOR SAMPLING AND ITS MAJOR APPLICATIONS When bulk products are bought and sold, two vital measures are required:- Quantity: how many tonnes or cubic metres, and Quality: chemical composition, physical properties, contaminants, etc. Total quantity for liquids can be measured accurately by tank gauging or flow meters. Quality is more difficult to measure. The quality of a whole batch cannot be measured directly. It can only be estimated by taking one or more samples and measuring the quality of these. The samples taken must be representative that is they must have the same properties as the material they represent. The number of samples taken to get a reliable measure of batch quality depends on the nature of the product and the operations involved. This course deals with the oil industry and primarily with sampling liquids. Oil is one of the world s largest industries. The quantity and monetary value of the oil and gas it handles is vast. Apart for the buyers and sellers, governments also have a major interest in its transactions because of tax revenues. Sampling, either manual or automatic, is a vital activity for every oil movement. More than 90% of automatic samplers worldwide are used to sample crude oils and other hydrocarbon liquids, for the determination of the water content. The financial importance of water measurement is easy to see. Crude oil is typically US$ 150 per cubic meter. Water has no value yet incurs costs to transport and process. A typical parcel of crude oil (one large tank ship) may be 100,000 cubic metres worth US$ 15 million. Just 0.1% by volume (v/v) of water is equivalent to US$ 15,000. Water contents in crude oils are typically less than 0.5% v/v but can be much higher. In one enquiry, the water content from an offshore field producing 60,000m 3 / day was reported to be between 35% and 85% v/v. The actual figure is unknown as the systems do not appear to get representative samples despite taking all precautions. The mixture was one of oil in water. 3

4 ORIGINS AND DEVELOPMENT OF AUTOMATIC SAMPLING The dominant application for automatic sampling is for wet crude oils. Oil and water do not mix. Water is heavier than oil and will sink to the bottom of tanks and pipelines unless the two phases are constantly being mixed. The whole technology of automatic sampling developed from the need to extract representative samples of wet crude oils flowing through pipelines. The International Standard ISO 3171 was produced for this application The first automatic samplers appeared about 35 years ago. Maurer and Jiskoot samplers were launched in the early 80 s Before ca. 1970, automatic sampling for crude oils was not necessary. Oil from the major onshore fields went into large tanks before shipment which allowed water to settle out and be drained off. The water content of oil received at refineries was typically < 0.05%. This was again held in local storage tanks so any water picked up during shipment could settle out. Crude oil then was only about US$ 10 per tonne so the price difference between oil and water was small. Oil quality was assessed from manual samples taken from the storage tanks. After 1970, the situation changed. The price of oil rose dramatically. More oil was produced from offshore fields with higher water content. More oil was sent through pipelines without using storage tanks, so water was not removed and manual sampling was not possible. Consequently there was an increasing demand to take samples from pipelines. Automatic samplers were developed to meet this need. Various methods were used taking either intermittent grabs or a continuous flow of sample. BP developed its first pipeline sampler in 1973 which was manufactured by Jiskoot and used on many North Sea platforms. They were primitive by today s standards but better than nothing. True Cut (later Cliff Mock) offered a sampler designed for clean oils, but it was considered unsuitable for crude oils. Around 1980, Welker (USA) developed its air-operated grab sampler using the vanishing cup technique to capture the sample grabs. But it had serious deficiencies. So in 1981, BP developed the 3-step grab sampling technique currently being made by Wellhead Metering Systems, Jiskoot, Doedijns and Maurer. Crude oils transported by sea are stabilised before shipment to remove volatile components. Samples can be collected in vented cans. From ca some crude oils e.g. from North Sea platforms, contained dissolved gases and therefore a high vapour pressure. To handle these HVP oils required special techniques, briefly described on pages 8 and 10. 4

5 WATER PROFILE IN PIPELINES MIXING TYPES OF MIXERS When a mixture of oil and water flows through a horizontal pipeline, the water will sink to the bottom unless the mixture is constantly agitated. To extract a representative sample, the flowing mixture must be well mixed so that the following two conditions are achieved:- 1. The pipeline mixture is homogeneous that is, the water content is uniform across the whole pipe section, and 2. The water is well dispersed that is, the water droplets are small compared with the opening in the sample probe. If these two conditions are not satisfied within ISO 3171 limits then the sample may not be representative. Users should check the water content profile at the sampling location to ensure conformity at all pipeline conditions. This is not difficult but is rarely done. Mixing oil and water requires energy. In most applications, mixing depends on having a high enough flow velocity, > 2.5 m/s such that natural flow turbulence does the mixing. Pipe fittings such as tees, elbows, Flowmeters create extra turbulence. The mixing energy is provided by the pipeline pumps. If the flow rate is too low for mixing by natural turbulence, then some additional artificial mixing should be installed. This is not common because of cost. Three types of artificial mixing are available:- The least cost is the static mixer which is a section of shaped vanes which tumble the oil in different directions. But this type is flow rate dependent and its pressure drop increases with (flow rate) 2. Special designs with variable geometry to maintain a constant pressure drop were developed and some were used at Maasflakte Terminal near Rotterdam on large diameter crude oil import lines. Another type is an electrically driven rotary mixer. It has low DP and is not affected by flow rate. It can be turned off when not required, but it can be expensive. The other technique is jet mixing which extracts a portion of the pipeline flow and pumps it back as a high velocity jet. The prime attractions of jet mixing are that it can be installed on a live pipeline and it does not obstruct the pipeline. The first successful Jet mixing head was supplied by Maurer. Jiskoot has to date exclusively exploited the Jet-Mixer System. 5

6 SAMPLING PROBES DESIGN AND LOCATION When the pipeline mixing conditions are satisfactory, the next task is to extract a representative sample. If the water profile was 100% uniform, a sample taken from any location across the pipe section would be the same. In practice, complete uniformity is difficult to achieve. ISO 3171 accepts a variation of +/- 5% but specifies from where in the section the sample should be taken, to minimise errors. In effect, the sample must be taken from on, or just below, the pipeline centre. Probes to ISO 3171 have only one sample entry whereas probes to the Russian GOST standard may have multiple entries (3 or 5) to mitigate the effects of nonuniform water profiles. MIL/FMA has supplied many GOST-pattern probes. Probe designs must meet two sets of criteria analytical and mechanical which often conflict. The final design is often a compromise. First and foremost, the probe must be suitable for the pipeline pressure and flow velocity and meet the users piping standards. For long probes and/or high velocities, the most difficult problem is to withstand the vibration caused by oil flow across the probe stem which induces vortex shedding. All probes have to be custom-designed to suit the application. The ideal probe should have a long forward-facing pitot and smooth flow path into the stem, to minimise disturbances to the flow pattern. In practice, the entry nozzle size provided by the customer allows only a short pitot or none at all. The sample should be taken isokinetically that is the velocity of sample entering the probe should be the same as the pipeline velocity, within certain limits. In many cases, this conflicts with the minimum velocity required in the probe stem and fast loop piping to avoid water drop-out. Probes may be fixed or retractable. Fixed probes are cheaper and pose fewer design problems. R. Maurer designed fixed probes for fast-loop systems because they can have a sample return connection. Retractable probes on high pressure systems (with thick flanges) and/or installed through double block and bleed valves often need very long stems which aggravates vibration problems. For In-the-Line (ITL) Samplers, the sampler is also the probe and must be retractable for servicing. WMS probes are normally made from 316 stainless steel, though mounting flanges and seal housings may be carbon steel to minimise cost. Samplers with all wetted parts in Duplex stainless (22% chrome) have also been manufactured. 6

7 PRINCIPLES OF GRAB SAMPLING TIME AND FLOW PROPORTIONING All modern automatic samplers are fixed volume grab samplers. Grab samplers are just reciprocating positive displacement pumps with special features to cope with wet and dirty oils. The sampling mechanism extracts a large number of small grabs from the flowing oil and pumps these grabs to a sample receiver. The total volume collected should be a representative sample of the oil parcel. The grab volume for the most manufactured samplers is normally 1 ml but can be 0.5 or 2 ml. It is defined by the bore and stroke of the sample chamber. The number of grabs depends on the application. In theory, increasing the number of grabs improves representativity but the number is limited by practical factors. For stabilised oils which can be collected in vented cans, the number is typically 5,000 to 10,000 to collect 5 to10 litres. For high vapour pressure (HVP) oils which must be collected in high pressure receivers, the number is typically 1,000 (i.e. 1 litre) for a daily sample and 3,000 for a weekly or monthly sample. Each grab should representative of the same volume of pipeline flow. As an example, if 1,000 grabs are taken from a cargo of 1,000 m 3 then each grab must represent 1 m 3. Note that the collected sample is only 1 millionth part of the cargo. If the flow rate is constant for the whole period of sampling, grabs can be taken at fixed time intervals. Using the above figures, if the flow rate was constant at 1 m 3 per minute then one grab would be taken each minute for 1,000 minutes. This mode of operation is called time proportional sampling. This mode may be used if the flow rate varies by no more than +/- 5% about the mean for the sampling period. If the flow rate can change during the sampling period, then the grab rate (frequency) must also be changed automatically in proportion to the flow rate. This mode of operation is called flow proportional sampling. In past years, time proportional operation was commonly used because it required only simple equipment, e.g. a cam-operated timer. Nowadays most samplers are controlled by computers and flow proportional operation is almost universal. Whichever mode is used does not affect the design of the sampling equipment. The signal which drives the sampler is normally provided by the customer from a central computer. 7

8 TYPES OF GRAB SAMPLER There are two distinct types of grab sampler and WMS manufacturers both types:- In the Line Samplers (ITL) are so called because the sampling mechanism is located inside the pipeline. The sampling mechanism is at the tip of a 1 OD stainless steel stem which is inserted into the pipeline through a ball valve mounted on a pipeline nozzle. Pipeline oil flows continuously through an inlet pitot tube and through the sampling chamber. In theory the operation is isokinetic, i.e. flow velocity through the chamber is the same as in the pipeline. When the sampler motor is actuated, a 1 ml volume of this flowing oil is first captured in the chamber then pumped up the probe stem and out to the receiver. ITL Samplers must be retractable for servicing, e.g. to replace worn seals. The WMS ITL Sampler has an integral mechanical retractor using twin lead screws. This allows the sampler to be inserted and withdrawn with the pipeline at full pressure and has other operating benefits. The stem slides through a seal housing. The Jiskoot ITL Sampler has a detachable retractor using hydraulic pistons energised by a hand pump. Cell Samplers also known as Fast Loop Samplers use the same sampling mechanism as ITL Samplers but are mounted outside the pipeline in a sample loop. A flow of oil is taken from a (fixed) probe in the pipeline, pumped round the loop though the Cell Sampler and back to the pipeline. The main advantages of a cell sampler are that it is readily accessible for servicing and can be more conveniently sited away from the pipeline. The fast loop sample may be more representative and other sensors (e.g. density and viscosity) can be mounted in the fast loop. Some users/contractors favour ITL systems on the basis of lower initial costs. 8

9 TYPES OF SAMPLE RECEIVER The sample grabs pumped by the sampler are collected in a sample receiver. The receiver must be able to contain the sample for days or weeks at a time without any change in the composition of the sample. ISO 3171 recommends that samples should be transported for testing in the primary receiver, i.e. in the receiver used to collect the sample. So receivers must be robust, easy to handle and to transport, which limits the size and weight when filled.. There are two basic types of receiver, depending on the nature of the sample. Fixed (or Constant) Volume Receivers: As the name indicates, these receivers have a fixed volume determined by the vessel dimensions. They are used for liquid samples having a low vapour pressure which can be stored and handled at nominally atmospheric pressure without losing light ends The type supplied for stabilised crude oils are commonly known as atmospheric cans. Typical capacities are 5, 10 and 15 litres. They are made in light gauge stainless steel, domed top and bottom to minimise water traps and a large access port with lid for easy cleaning. They can withstand pressures of a few Barg but are not classed as pressure vessels. They have connections for filling, a dip tube for use with lab mixers and facilities for level measurement. Variable Volume Receivers (or Piston Receivers): These are pressure vessels used for high vapour pressure liquids such as HVP crude oils, condensate and LPG. They are fitted with an internal piston loaded on the outboard side by inert gas such as argon or nitrogen. The loading pressure must be kept higher than the sample s vapour pressure, to maintain the sample in single phase, i.e. fully liquid. The piston may have a shaft or be magnetic, to allow indication of the piston position, hence contents. Sample is collected beneath the piston. The variable volume is the volume occupied by the sample, which ranges from zero to full. These receivers are also known as Constant Pressure (CP) Receivers because the gas loading pressure is normally kept constant. Piston receivers are expensive to manufacture and large volume pressure vessels can be very heavy. The market is dominated by Proserv NV and Welker, for both liquids and gases. Popular sizes for liquids are 1 and 4 litres. In the past, Bob Maurer designed and manufactured a wide range of piston receivers with pressure ratings up to ca. 200 Barg and volumes ranging from 50 ml to 10 litres. A special type of piston receiver unique to Maurer is the Piston Internal Mixing Receiver or PIMR. This comprises two piston receivers back-to-back, connected by some small orifices. It was developed by BP for the collection and mixing of HVP crude oil without losing light ends. Many PIMRs are used in North Sea operations. 9

10 COMMENTS ON SAMPLING SYSTEMS A basic sampling system comprises the probe, fast loop, samplers and receivers and the piping, connections and enclosures needed for safe and reliable operation. The system may include indicators for loop flow and receiver contents, multiple receivers with change-over facilities and other property sensors such as density. Isokinetic sampling and fast loop flow. The loop velocity ahead of the sampler must be high enough to prevent water drop-out. It should be >2.5 m/s but >2 m/s may be adequate. The flow to achieve that velocity will depend on the piping bore. The system designer aims to minimise the flow rate to use the smallest pump. At the same time he/she may want to use large bore piping to minimise pressure drop. System design is often a compromise but the over-riding factor must be to achieve a high enough velocity for good sampling. Other sensors e.g. for density and viscosity may be installed in the fast loop and these have specified flow rates which must be accommodated. The loop flow is also the flow through the probe. Its dimensions may be sized to tolerate flow-induced vibrations. Also the sample velocity at probe entry should be the same within specified limits as the pipeline velocity i.e. isokinetic sampling. So the probe dimensions cannot easily be changed to suit flow loop conditions. Samples containing sediments. Most crude contain some sediments, e.g. sand and grit. The total non-oil content is termed BS&W, i.e. Bottoms Sediments and Water. Measures to mix the oil and water will also mix the sediment. The ISO and ASTM standards do not permit using filters to protect loop devices because they change the sample make-up. Strainers may be acceptable. WMS samplers have elastomeric seals and can handle abrasive sediment. But other system components need to be carefully selected to handle sediment. So centrifugal pumps are preferred to gear pumps, for gritty oils. The sediment content might better be determined on a manual sample taken from the loop upstream of the strainers. Sampler outlet to receiver connection. This must be short, < 1 metre and must have a continuous downward slope to avoid water traps. This is not a problem for Cell Samplers where sampler and receiver are close together. But for ITL systems, mounting the receiver close to the sampler can be difficult. 10

11 SAMPLE HANDLING, MIXING, SUB-SAMPLING AND TESTING Having a receiver containing say 1 litre of representative sample does not measure anything. That sample has to be tested to get any information on oil quality. The receiver has to be taken to a laboratory, fully mixed and then a small but representative fraction taken for testing. The original 1 litre sample may represent 10,000 m 3 of cargo that is 1 part in The test for water content may be performed on < 1 ml of sample, only 1 part in These ratios highlight the importance of representative sampling at every stage. The mixture in the receiver may have been standing for days or weeks. Any water and sediment will have settled out. So the first step is to homogenise the sample using high energy mixing... Shaking or stirring is not capable of mixing oil and water to the degree required. For stabilised oils, a high-shear rotary mixer may be placed in the receiver. Or the oil can be circulated through a loop containing a small static mixing element. Proprietary packages are available for this duty. Maurer s PIMRs (Internal mixing receivers) were developed to collect HVP samples and to provide high energy mixing within the receiver, by shuttling the two pistons. Proserv homogenised the samples via an external pumping system When the mixture is homogenised, the next step is to take a sub-sample for testing. For stable oils, this is not a problem. But for HVP oils, the sub-sample must be taken from a pressurised system, to avoid losing volatile components. This can be done with a special hypodermic syringe designed to work at high pressures. The PIMRs have a facility to connect a pressure syringe and take a subsample during the mixing process, to reduce any risk of water drop-out. Finally, the actual testing. The recommended test method for water content is the Karl Fisher coulometric titration. The oldest method is by centrifuge but its accuracy is low. It is still much-favoured by some operators, especially in US companies because of its simplicity. An intermediate method is Dean and Stark distillation. The best test method is accurate to better than 0.01% water. The worst method s accuracy is not better than 0.1%. Water content is normally measured on daily samples because of its importance. Bulk oil properties like density and hydrocarbon make-up are performed on weekly and monthly samples. These tests require a larger volume of sample for testing. 11

12 CONTINUOUS MEASUREMENT OF WATER CONTENT The earlier sections of this note have indicated that automatic sampling is a complex and antiquated approach to measuring the water content of crude oils. It provides only one sample from each batch, it has many sources of error and the one test result is only produced long after the batch has gone In the oil and chemical industries and in many others, the properties of flowing gases and liquids and even solids are measured continuously using on-line analysers. Therefore you may ask if water content can be measured continuously and, if so, why is it not used in place of automatic sampling. Water content monitors are available. They have been available for at least 15 years. A provisional ASTM standard for water monitors was produced ca. 1999, but it was probably intended for non-fiscal duties. Water monitors operate by measuring changes in an electrical property (dielectric constant) of the mixture. The dielectric constant of oil is very low (ca. 2) and that of water is very high (ca. 80). The sample flows through a tubular cell with a central electrode and the monitor measures the electrical capacitance of the cell. The monitor is very simple and robust and suitable for high pressures. It is claimed to detect changes in water content of 0.01% and to be accurate to 0.05%. It can measure up to ca. 40% water. These claims might not apply in all situations. Why have water monitors not replaced samplers? A major reason is industry inertia. Automatic sampling is recognised and used worldwide for custody transfer and fiscal measurements. Replacing it with a different method would be very difficult. It is more likely that water monitors will be installed initially alongside conventional grab sampling, to have the benefits of both. This is already happening. Water monitors need a representative sample, so the same pipeline mixing and fast loop facilities as for automatic sampling are still required. The installed system costs to have both systems may not be substantially greater than for either one alone. The extra costs may be justified by the benefits of having real time data and two separate measurements of the water content. Finally, automatic sampling provides an actual average sample which can be retained for reference and could be tested by all parties. Monitors only provide an electronic record. It is interesting to observe that oil quantities are measured by flow meters which also provide only an electronic record. END 12