Effective interface control in multi-phase separators

Various measurement technologies have been used over the years to measure interface in a separation vessel. Interfaces vary from clean and well-defined, to a small rag layer or a large emulsion band, right through to multiple interfaces of oil, emulsion, water, solids and associated build-up. Different technologies have shown varying degrees of success, but rarely cope with all of the conditions that can occur. Doug Anderson of VEGA Controls Ltd asks how can this measurement be made more reliable?

‘Traditional’ technologies used for interface measurement have shown degrees of unreliability, especially with heavy emulsions and/or varying quality of crude inputs. There are three main reason:

1. They need consistent ‘measurement properties’ of the liquid phases they are measuring in order to function, e.g. electrical property or density/specific gravity. If it is not constant or stable, it introduces another (unwanted) ‘unknown’ into the measurement ‘equation’.
2. Measurement devices can also be affected by build-up or contamination from the process over a period of time.
3. As well the requirement for stability, most technologies for interface monitoring also require a minimum ‘measurement property’ differential between the upper and lower fluids to detect a difference (interface), otherwise they start to ‘guess’ or fail completely. The problem with emulsions is that they are not defined, they are diffuse in their nature and literally a ‘grey area’ to many interface detection technologies.

Often the end user may be unaware of the characteristics of the process/interface they are trying to measure, especially if there is no possibility for the use of externally mounted sight glasses on the vessel to try and ‘see inside’ the process. It therefore makes it doubly important to have a measurement technology that is proven and reliable in managing the throughput of any separation process.

Degrees of separation

Process vessels such as separators, treaters, free water knockout drums and desalters, can often be at the bottle neck of an operation. A continuous ‘clean and distinct’, well-behaved interface boundary is a rarity. This may only be achieved by longer residence times, (not always possible) or using large quantities of effective chemical ‘demulsifiers’ (expensive), running the process at higher operating temperatures or employing specialised state of the art separation technology (not always feasible).  Producers will often have a separation process that has some degree of an emulsion or rag layer involved, and this can be due to external factors – such as varying crude quality, its origin or process upsets further upstream. In these applications, monitoring this diffused layer between the upper and lower fluids is critical to maintain the balance between maximum throughput and the effectiveness of the separation process. But what to use? How effective is it? How much is needed?

A multi layered solution

When the measurement property difference used for making the interface measurement is too small or diffused within an emulsion layer, reliable control is not possible for most legacy technologies. One solution, however, uses the variability of this ‘measurement property’ as the actual basis of its measurement. This is achieved by an array of optimised nuclear gauging systems making multiple independent, density measurements across defined elevations in the vessel. By monitoring differences in density from one elevation point to the next, the operator has a fixed reference ‘density profile’ that monitors the dynamic changes in the separation process on a real-time basis. The detection of subtle density changes can verify the development and efficacy of a separation process, especially when subjected to multiple blended crudes of varying characteristics, slugs of water, and the various chemical treatments and process adjustments mentioned earlier. This ‘window’ into the process can enable the automatic control of water withdrawal and optimise chemical inputs.

This type of system is called a multi-point density array (MDA), and it is perfectly adapted for these difficult interface/heavy emulsion processes. By measuring changes at each fixed point, the operator can monitor whether the process is trending towards higher density (brine) or lower density (oil). If no clear interface exists, the operator still sees a gradual density decrease from the brine-detector point at the bottom, to the oil-detector point towards the top of the vessel. A total level can also be output, as well as monitoring for the build-up of sand and solids. This means independent SIL point level/safety trips can operate off the same dry-well source installation, using separate detectors for increased security. The system is not ‘unique’ in its operating principle, but it offers some unique gains for the user and process. It’s accurate, provides real-time data, is scalable, flexible, cost effective and robust. It is easy to understand, and relatively easy to install and operate.

Multi-density array architecture

The system employs a single inserted dry well carrying the nuclear sources, which can be safely withdrawn into a source holder to take the system offline, should vessel access be required. The discrete collimated (focused) sources provide optimum energy for performance, density and resolution.

Each source has a dedicated externally mounted high-accuracy scintillation based detector, so there is no crosstalk and no ‘averaging algorithms’. They are simple to set up and calibrate on a ‘set density’ of liquid, e.g. process water. As the outputs change from each of the multiple detectors in the array, the operator can determine the movement of the interface or the size of any emulsion layer inside the vessel. The density profile created by an MDA system provides actionable data that enables real-time process monitoring, leading to unit efficiencies that cannot be achieved by other measurement technologies.

System Components of a Multi Density Array

The number of sources, source holders, and detectors are scalable and the location of detectors are application-dependent, so you only buy and use what you need.

1. Long-life scintillation density detector with either aluminium or stainless steel housing
2. Source holder allows safe withdrawal of gamma sources to take the system offline
3. Sources in dry well with discrete collimated beam for maximum accuracy
4. Dry well carries the sources and is designed around the specific vessel and application requirements of the end user
5. Flexible detector mounting bracket system

System Outputs

Individual sources are matched with individual detectors to make independent density measurements in a horizontal plane. Each gauge outputs a 4 … 20 mA/HART output signal to the producer’s DCS, which is easy to manage and produce a process-control screen for operators. An example of the typical outputs is shown below:

Why consider an MDA type profiler system? The benefits to the end user are as follows:

• Density profile across the elevation span monitors changes in the separation process on a real-time basis, enabling automatic control and optimal throughput
• Each of the density gauges independently transmit their signal to the control room. No external algorithms on a separate computer are required. Easy for fault finding.
• Safe withdrawal of sources using extremely simple and robust retractable mechanisms. Relatively small vessel process connection.
• Reliable, even with process build up inside, through higher energy output and longer measurement paths – better performance and reduces outages for cleaning or recalibration
• System can also support extra independent units for total level and point level trips (SIL)
• Externally mounted SMART detectors are reliable by design, withstand higher process temperatures for low operating costs
• Fast calibration of sensors and system. Monitoring each fixed point permits trending optimising inputs and throughput of the process
• If it should happen, a single density detector outage does not compromise the system. If hardware exchange is needed, it is faster due to external mounting and back up of calibration.
• When no clear interface exists, the system shows valuable data to regain process control in the quickest time using optimal inputs.