Two-phase flow pattern identification using continuous hidden Markov model

Mahvash, Ali; Ross, Annie

doi:10.1016/j.ijmultiphaseflow.2007.08.006

Cited by 30 publications

(8 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the micro-optofluidic chip presented in [14], the advantage of the integration of micro-optical and microfluics components in one device is proved by taking the advantage of advanced signal analysis methodology to process the optical information and control the flow. Advantages can be also be envisaged for SoC applications due to the simplicity of managing optical signals, as it is proved by a wide literature on flow classification in microchannels [15,16].…”

Section: Introductionmentioning

confidence: 99%

Real-Time Detection of Slug Velocity in Microchannels

2020

View full text Add to dashboard Cite

Microfluidics processes play a central role in the design of portable devices for biological and chemical samples analysis. The bottleneck in this technological evolution is the lack of low cost detection systems and control strategies easily adaptable in different operative conditions, able to guarantee the processes reproducibility and reliability, and suitable for on-chip applications. In this work, a methodology for velocity detection of two-phase flow is presented in microchannels. The approach presented is based on a low-cost optical signals monitoring setup. The slug flow generated by the interaction of two immiscible fluids {air and water} in two microchannels was investigated. To verify the reliability of the detection systems, the flow nonlinearity was enhanced by using curved geometries and microchannel diameter greater than 100 μ m. The optical signals were analyzed by using an approach in a time domain, to extract the slug velocity, and one in the frequency domain, to compute the slug frequency. It was possible to distinguish the water and air slugs velocity and frequency. A relation between these two parameters was also numerically established. The results obtained represent an important step in the design of non-invasive, low-cost portable systems for micro-flow analysis, in order to prove that the developed methodology was implemented to realize a platform, easy to be integrated in a System-on-a-Chip, for the real-time slug flow velocity detection. The platform performances were successfully validated in different operative conditions.

show abstract

Section: Introductionmentioning

confidence: 99%

Real-Time Detection of Slug Velocity in Microchannels

2020

View full text Add to dashboard Cite

show abstract

“…Commonly used methods for characterizing the flow regimes are direct observation, analysis of physical time‐average quantity, and characteristic analysis of fluctuation signals. The observation method is simple but subjective and limited to laboratory‐scale fluidized beds. As suggested by the name, the analysis method of physical time‐average quantity is mainly used to relate the flow dynamic characteristics to the flow regimes, such as axis or radial voidage and particle velocity .…”

Section: Introductionmentioning

confidence: 99%

Characterization of flow regimes in fluidized beds by information entropy analysis of pressure fluctuations

2016

View full text Add to dashboard Cite

Flow regimes in a gas‐solid cylindrical fluidized bed of 150 mm (i.d.) × 1.2 m (high) with three different sand masses (1.5, 3, and 4.5 kg) were studied via information entropy analysis of pressure fluctuations. Three classes of methods of information entropy were adopted to characterize the flow regimes. It is shown that the first‐class dimensional methods are suitable for illustrating the distinct characteristic of the flow regimes under different operating conditions, while the other two dimensionless methods are appropriate for finding the similarities or common rules. Tsallis entropy and Renyi entropy do not provide more valuable information than Shannon entropy. Most importantly, the Shannon entropy method is a good choice for characterizing the flow regimes but is weak at revealing the transition velocity (Uc) clearly. This deficiency can be made up by the method of component Shannon entropy. Shannon entropy is partitioned into 4 component Shannon entropies (named CH1CH4), which may be related with a 4‐stage fluidization process assumption. A rule is found that the transition of flow regimes transition from bubbling bed to turbulent bed takes place when the contribution proportion of CH3 is exceeded by that of CH1, and it brings new enlightenment on interpretation of the transition mechanism.

show abstract

“…The fluctuating signals about flow regime have many feature extraction methods, such as time series analysis, wavelet, time-frequency joint analysis, fuzzy, chaos and fractal by Tutu 1984;Matsui (1986); Wambsganss & Jendrzejczyk (1991); Cai, Wambsganss & Jendrzejczyk (1996),; and Langford, Beasley & Ochterbeck (1998). The artificial neural network has being developed to a conventional method to recognize the flow regime by Mi, Ishii &Tsoukalas (1998); Monji & Matsu (1998); Wu & Zhou (2001); and Mahvash & Ross (2008).…”

Section: Introductionmentioning

confidence: 99%

Online recognition of the multiphase flow regime and study of slug flow in pipeline

Guo

Bai

Zhao

et al. 2009

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

Multiphase flow is the phenomenon existing widely in nature, daily life, as well as petroleum and chemical engineering industrial fields. The interface structure among multiphase and their movement are complicated, which distribute random and heterogeneously in the spatial and temporal scales and have multivalue of the flow structure and state [1] . Flow regime is defined as the macro feature about the multiphase interface structure and its distribution, which is an important feature to describe multiphase flow. The energy and mass transport mechanism differ much for each flow regimes. It is necessary to solve the flow regime recognition to get a clear understanding of the physical phenomena and their mechanism of multiphase flow. And the flow regime is one of the main factors affecting the online measurement accuracy of phase fraction, flow rate and other phase parameters. Therefore, it is of great scientific and technological importance to develop new principles and methods of multiphase flow regime online recognition, and of great industrial background.In this paper, the key reasons that the present method cannot be used to solve the industrial multiphase flow pattern recognition are clarified firstly. Then the prerequisite to realize the online recognition of multiphase flow regime is analyzed, and the recognition rules for partial flow pattern are obtained based on the massive experimental data. The standard templates for every flow regime feature are calculated with self-organization cluster algorithm. The multi-sensor data fusion method is proposed to realize the online recognition of multiphase flow regime with the pressure and differential pressure signals, which overcomes the severe influence of fluid flow velocity and the oil fraction on the recognition. The online recognition method is tested in the practice, which has less than 10 percent measurement error. The method takes advantages of high confidence, good fault tolerance and less requirement of single sensor performance. Among various flow patterns of gas-liquid flow, slug flow occurs frequently in the petroleum, chemical, civil and nuclear industries. In the offshore oil and gas field, the maximum slug length and its statistical distribution are very important for the design of separator and downstream processing facility at steady state operations. However transient conditions may be encountered in the production, such as operational upsets, start-up, shut-down, pigging and blowdown, which are key operational and safety issues related to oil field development. So it is necessary to have an understanding the flow parameters under transient conditions. In this paper, the evolution of slug length along a horizontal pipe in gas-liquid flow is also studied in details and then an

show abstract

Two-phase flow pattern identification using continuous hidden Markov model

Cited by 30 publications

References 17 publications

Real-Time Detection of Slug Velocity in Microchannels

Real-Time Detection of Slug Velocity in Microchannels

Characterization of flow regimes in fluidized beds by information entropy analysis of pressure fluctuations

Online recognition of the multiphase flow regime and study of slug flow in pipeline

Contact Info

Product

Resources

About