Microwaves analyse multi-phase flows in pipes

A novel sensing system uses low power microwaves from a potentially low cost source to determine proportions of gas, water and oil inside a pipeline. Presently on trial with a major oil company, it works with a neural network to measure amounts to single percentage accuracies. The only device we are aware of which can analyse mixtures containing three of more different phases, in a completely non contact manner, it has great potential to solve a large number of process monitoring problems, and may even provide the means to apply non obtrusive security scans at entrances to public places and transport systems. The idea is mainly the brainchild of Dr Andrew Shaw and Professor Ahmed Al-Shamma'a, who formerly worked in the University of Liverpool's Department of Electrical and Engineering, headed by Professor Jim Lucas, but is now to be found in the Radio Frequency and Microwave Applications Research Group at Liverpool John Moore's University The task set was to find a way of metering multiphase wet gas for the oil and gas industries. The flow emerging from a typical undersea well not only contains oil and gas but also water, salt and sand. The proportions change as time passes and also as pressure falls as the mix flows up towards the surface of the sea. Companies often share docking platforms and tankers, so they need to know exactly how much oil or gas their field has contributed to a tanker load, as opposed to a neighbouring field operated by a competitor. Furthermore, unwanted substances have to removed and good online sensing is desirable for process control. "There are so many sensors around but none can cope with three or more phases", Professor Al-Shamma'a exclaimed when presenting the results of his work at the Institute of Physics conference on "Sensors and their Applications XIII", just held in Chatham on their new University of Greenwich campus. The basic idea is to construct a suitable cavity round the outside of the pipeline, and pass microwaves through it from one loop antenna to another one on the opposite side. The dielectric constants of the contents of the pipe vary with frequency, so sweeping the frequency of the transmitter produces a received signal whose magnitude and phase shift also varies with frequency. Because pipelines are fairly large, the wavelengths of signals appropriate to the dimensions of the exterior cavity are relatively long, with frequencies in the range 100MHz to 350MHz, which means they can be generated, detected and amplified using low cost components. Because the relationship between contents and output signal as a function of frequency is extremely complicated, content composition is deduced by a neural network. This was trained first with pure contents of oil, water and gas and then with mixtures. Extensive modelling using Ansoft's High Frequency Structure Simulator (HFSS) software led to a cavity and pipe configuration which worked well in the laboratory, leading onto further trials at the National Engineering Laboratory in East Kilbride. Parts were made by Solartron ISA and the tests conducted on a real piece of gas/oil pipe, pressurised at up to 150 bar with the outer cavity first filled with water and then with oil. Output data was compared with that from a gamma ray densitometer, the currently favoured technology to analyse two phase flows inside pipes. This has led onto a prototype system with an input signal scanned from 150MHz to 300MHz using a voltage controlled oscillator. Power is only a few tens of microwatts. Output is passed through a 12-bit analogue to digital converter and passed to a PC running the neural network and performing the analysis. The prototype system is now on trial with BP. Despite the simplicity of the system and the low power levels, it has been found able to detect down to 1ppm oil in water, well below the 10ppm permitted limit for water to be discharged. It has also been found able to cope with even more complicated multiphase flows, such as those containing sand as well as water, gas and oil. Accuracy in the case of detecting small percentages of oil or water in a gas pipeline is found to be as high as 99%, as compared with 50% using other technologies. It thus looks to be an eminently suitable technology for the process industries and possibly the food industries also, which are constantly concerned to maintain the integrity of their products. It may also have many other applications thanks to the potentially low cost of the components and the low power levels. There has for example, long been an established need for some system or systems that could pick up the presence of drugs, explosives, or other less dramatic but equally problematic contraband, such as banned animals and quantities of foodstuffs, over and above those intended for personal use as well as gas cylinders and other undesirable cargo being carried onto public transport. While it is extremely unlikely that any system could possible detect everything of interest, we did ask at the conference whether the system could be made to detect a rucksack full of explosives and were advised that it should be able to make it do this. Despite constant claims, mainly made by US companies and research teams that they have the basis of a technology to do this, none, to our knowledge, are anything like as near to being a manufacturable system. A patent has been applied for by Liverpool University where much of the original research work was undertaken and has currently taken up by Solartron ISA. The research work was supported by the DTI under the Basic Technology programme managed by NEL and supported by Solartron ISA and BP. Email Professor Ahmad Al Shamma'a Eureka says: This could be the breakthrough in non contact analysis of mixtures of substances going down pipes and ducts that many researchers have been looking for Pointers * System can detect and measure the composition of flows containing three or more phases * Frequency range is 100MHz to 350MHz and power levels are only tens of microwatts * Measurement limits are in some cases down to 1 ppm and measurement accuracy up to 99%.