contact    site map  
    polski   english

Process Tomography

 

"tomoKIS" research team members
Prof. D. Sankowski, D.Sc. eng.
W. Mosorow, D.Sc.eng., prof.PŁ
L. Babout, D.Sc.
R. Banasiak, Ph.D. eng.
Z. Chaniecki, Ph.D. eng.
K. Grudzień, Ph.D. eng.
M. Janaszewski
, Ph.D. eng.
J. Nowakowski, Ph.D. eng.
A. Romanowski, Ph.D. eng.
R. Wajman, Ph.D. eng.
P. Fiderek, M.Sc. eng.
T. Jaworski, M.Sc. eng.

Research in the field of process tomography is developed under the supervision of prof. D. Sankowski with the inspiration and the substantial assistance of prof. A. Plšskowskiego from Warsaw. The success, for research on process tomography, the institut owes to the deceased in 2006, Prof. T. Dyakowski. Thanks to its high activity the institut has won worldwide acclaim and high academic positions in the field of capacitive tomography.

In August 2008, and also in September 2004, the institut has organized two International Symposiums on these issues, attended by more than 80 scientists from China, Czech Republic, Finland, Greece, Japan, Lithuania, Germany, Norway, Polish, Ukraine and the UK.

Prof. D. Sankowskiego is a member of the International Advisory Committees on Congresses and was chosen to be the Communication Director of the International Society for Industrial Process Tomography ISIPT.

 

The Tom Dyakowski Process Tomography Laboratory

Since October 28th, 2008 in the Institut of Applied Computer Science is available process tomography laboratory. It is one of the most modern laboratories in the world. The laboratory has name of Prof. Dyakowski Thomas - who died prematurely in 2006, co-founder of the process tomography school in Lodz, an employee of the Technical University and the University of Manchester.

Tablica Pamiatkowa ku pamięci profesora Tomasza Dyakowskiego

The laboratory includes the facilities for non-invasive physico-chemical phenomena research occurring in the multiphase flows. In order to ensure that research conditions are similar to that during the real industrial processes, the facilities are designed and built in a semi-industrial scale. Constructed facilities are used for research of:

  • two phase gas - liquid flows,
  • pneumatic transport of loose materials,
  • gravitational flows and phenomena while silo discharging,
  • liquids mixtures.

 

The pictures of research facilities in the laboratory

The ECT system applied for non-invasive flow diagnosis

In the laboratory the tomography and other devices can be found:

  • 16 channels capacitance tomography made by ECT Instruments Ltd. Thank to its speed it is mostly applied to the flow diagnosis and monitoring,


     
  • Two 32 channels capacitance tomography construct by the research team (dr R. Szabatin) from Institute of  Radioelectronics at Warsaw University of Technology. The main feature of that devices is the possibility of freely gains configuration for each of the measurement channels. Therefore they are mostly suitable for three dimensional tomography research,

    foto tomograf ET3
  • 32 channels electrical tomography build in Institut of Applied Computer Science in a frame of DENIDIA  project. This device is applied to the three dimensional monitoring as well but also it supports the research on the rotatable sensors thanks to its possibility of measurement synchronization. In addition this tomograph can measure using  both capacitance and resistance technique,

  • and the measurement equipment by National Instruments, Rigol i Agilent.

The IT section of the tomography system is composed of a high performance PC's:

  1. tomoKIS Server: 2 x Intel Xeon Quad 2.33Ghz, 16 GB RAM, 1x GPU ATI Radeon 4850            
    (theroretical computational power: 1 TFlop)
  2. TESLA Server: Intel i7 2.8 Ghz, 12 GB RAM, 2 GPU x ATI Radeon 5970, serwer 4 x GPU TESLA S1070-400    
    (theroretical computational power: 13,3 TFlops)
  3. Radeon Server: AMD Phenom II 955 Black Edition 3.2Ghz, 8 GB RAM, 4 x ATI Radeon 5970           
    (theoretical computational power: 18,4 TFlops)

 

Pictures of computation IT system

The computational potential of the Radeon Server is demonstrated on next video. The Radeon based system reaches over 280000 score points using computing benchmark (see video). To compare: the Intel i7 2.8 GHz Quad Core CPU reaches 3600 scores for the same computing test

Such a set of measurement and computing systems allows a visualization of both 2D and 3D study of the industrial process es in the real time. The laboratory facilities have resulted in both the scientific works published in the journals from the Thomson Reuters Master Journal List as well as many prizes at national and international exhibitions (follow this link).

The laboratory plays a significant role in the development, testing and implementation of non-invasive imaging systems for industrial applications. The aim of this project is to develop a modern non-invasive techniques for the automatic control of the industrial processes in terms of quality and optimize production.

DENIDIA Project News

Since October 2006, the institut has been coordinating a 1 M€ Marie Curie-Transfer of Knowledge project, called DENIDIA (Development of Excellence in Non Invasive Diagnostic system for Industrial and research Applications). The project is in its 3rd year and has already achieved some important results in the field of process tomography, thanks to the recruitment of 3 Post Docs, 1 visiting Professor and the secondment of 3 permanent researchers to selected partners. These results include:

  • development of flexible AC 32-channel ECT system that incorporates TCP/IP protocol for communication between sensor and tomograph,
  • new image reconstruction algorithm for measurements from 3D ECT,
  • development of Rotatable ECT sensor for improved resolution,
  • development of dual modality system combining Gamma-ray with ECT measurements,
  • new advanced 3D image processing algorithms applied to materials science problems.

example of non-linear 3D image reconstruction

Example of non-linear 3D image reconstruction.

 

 
General view of the rotatable sensor. Result of image reconstruction of a square-like cross section object.(a) Image from a classical 16-electrode sensor, (b) Image from rotatable sensor with 4 angle positions.

 

hole filling - 3D image processing algorithm application
Example of 3D image processing algorithm applied to materials science problem, called hole filling.

More details about the previous examples of research achievements in the frame of the DENIDIA project can be found http://www.kis.p.lodz.pl/denidia.html

Introduction to Process Tomography
The word "tomography" is derived from the Greek tomos (slice) and graphia (description) and usually the associated technique considers the image reconstruction of a cross-section of objects. The first application was carried out in medicine more than 30 years ago, but it is during the last decade that the technique widened to industrial and research studies. While techniques like X-ray computed tomography are usually performed on static objects, another range of non-destructive techniques (such as Electrical Capacitance Tomography, Optical Tomography, g-ray tomography), called generally process tomography, have also been developed in order to study fast technological processes without disturbing their behaviour. The first application of process tomography was developed at the University of Manchester in 1980 to characterise multi-phase flow process in pipeline during oil extraction. The technique is mainly used to characterise density gradients and inhomogeneity, flow velocity. The main advantages of the process tomography are to enable fast acquisition, real-time analysis thanks to fast image reconstruction algorithms, and low cost development (ie. Electrical Capacitance Tomography (ECT), Optical Tomography). However, its main drawbacks are to present a low spatial resolution (due to the sensor sizes and ill-posedness of reconstruction methods due to so-called soft field effect), and to usually provide only two-dimensional measurements, as opposed to classical X-ray tomography. Research study at the Institut's level has already focused to develop 3D-ECT and are presented in the following section.

The research program on process tomography has been initiated in the Institut by Prof. Dominik Sankowski and Prof. Tomasz Dyakowski, in collaboration with Prof. Andrzej Pląskowski from the Warsaw University of Technology. The research carried out in the Institut of Applied Computer Science mainly focuses on

 

  1. development of new ECT designs (twin-plane systems, 3D sensors),
  2. new image reconstruction algorithms to increase spatial resolution, combine different modalities (for instance, g-ray tomography and ECT, Optical Tomography and ECT) and allow 3D measurements,
  3. computing methods to calculate velocity with twin-plane system,
  4. data interpretation of different powder flow phenomena (flow in pneumatic conveying and silos, hopper flow, flow of friable materials).

Most of the studies involve collaborations with national institutes:

  • Division of Nuclear and Medical Electronics, Warsaw University of Technology,
  • Department of Fundamentals of Building and Material Engineering, Technical University of Gdansk,
  • Department of Heating Technique and Industrial Apparatus, Opole Technical University,

and international institutes:

  • Department of Chemical Engineering and Analytical Science, University of Manchester (UK),
  • Department of Statistics, University of Leeds (UK),
  • Department of Physics and technology, University of Bergen (Norway)
  • Industrial Technology Division, Malaysian Institute for Nuclear Technology Research (Malaysia).

Most of the research studies are supported by national and international grants (Polish State Comity of Science, EPSRC, Royal Society, IAEA).

The Institut is also strongly involved in dissemination of process tomography science. It organised the 3rd International Symposium on Process Tomography in Poland (Lodz, 9-10 September 2004). The conference welcomed 40 international speakers from China, Greece, Japan, Norway, Ukraine, United Kingdom and Poland. Prof. Dominik Sankowski was also a member of Advising Committees of 3rd and 4th World Congresses on Industrial Process Tomography in Banff (Canada, 2003) and Aizu (Japan, 2005). The Institut is also under negotiation to organise the 6th world congress in ŁódŸ in 2009/2010.

Research programmes

ECT equipment and designs

Involvements: Prof. D. Sankowski, Prof. T. Dyakowski, M.Sc. K. Grudzień, M.Sc. A. Romanowski, M.Sc. Z. Chaniecki, M.Sc. R. Banasiak.

The Institut is currently equipped with different modalities of systems (1x16 electrodes, 1x12 electrodes, 2x8 electrodes) with an acquisition speed of 30 frame/s. The systems have been developed in collaboration with the Warsaw University of Technology (Dr R. Szabatin). The system interface can control 32 channels. M.Sc. K. Grudzień and M.Sc. A. Romanowski are responsible for the maintenance and optimisation of the different kits.

M.Sc. Chaniecki has also designed, in collaboration with the Technical University of Gdansk and the Warsaw University of Technology, a dedicated twin-plane ECT to measure flow process of friable materials in silo. The silo, made of Plexiglas, is 2 m high with an outer diameter of 0.2 m. Each plane is composed of 12 sensors. The two sensors are placed on the outer surface of the wall at the distances of 0.5 m and 1.5 m from the top.

In the near future, more systems devoted to 2D and 3D- ECT will be developed (1x32 for large vessel, 2x12 and 3x8 for 3D-ECT and velocity measurements) with a faster acquisition (up to 200 frame/s). The work on 3D-ECT design has been initiated by M.Sc. R. Banasiak, who has also developed dedicated data reconstruction algorithm (see the following section). What is the difference between 2D and 3D measurements? The major difference lies in the fact that measurements are completely obtained from three-dimensional space using multi-plane and multi-electrode sensor. The electrodes are located around 3D region of interest and in comparison with 2D tomography, they are not placed on single plane only. In fact electrodes placement is quite flexible and can be adapted to application needs. However, the sensors, measuring unit and measurement strategy have to be balanced, because the difference in capacitance value between the closest and furthest electrodes is large. It is obvious that in many cases 3D image provides much more information about process in compare to 2D image form classical ECT systems.

Image reconstruction algorithms

Involvements: Prof. D. Sankowski, Prof. T. Dyakowski, Dr W. Mosorow, M.Sc Ł. Mazurkiewicz, M.Sc. R. Banasiak, M.Sc. R. Wajman, M.Sc. S. Lewandowski.

Because of the non-linearity of the electric field measurements, the methods of image reconstruction always face a problem of ill-posedness. The best solutions of the reconstruction process usually correspond to solutions of forward problem (determination of measurement vector from assumed material parameter distribution - usually performed thanks to space meshing and FEM calculation), and/or backward problem (reconstruction of the material parameter distribution - in the case of ECT, the electrical permittivity - from measurement vector - usually performed thanks to Back Projection Algorithm (BPA) or Algebraic Reconstruction Techniques (ART)). The choices of the reconstruction strategy and technique are usually linked to the operator's needs. Most of work on the 2D reconstruction process has been carried out by M.Sc. Ł Mazurkiewicz and M.Sc. R. Wajman. Commonly used image reconstruction methods are not accurate for ECT systems. The reconstructed images are not satisfying in term of contrast and there is a lack of clear shape edges between the components of the studied process. During this research programme, a new image reconstruction method for ECT system was developed. Its innovation lies on two elements: the method for sensitivity matrix calculation along electrical field lines and the algorithm for nonlinear image reconstruction. The obtained results have proved that the ECT system for the real-time process visualization can be built. The new solutions (algorithms) can, in a significant way, accelerate the convergence of the image reconstruction process and simultaneously improve the final image quality. The mentioned method has already been applied to the identification of a two-phase flow structure, in collaboration with the Technical University of Opole.

Dr W. Mosorow has proposed reconstruction algorithms which take into account gamma-ray tomograph systematic errors which are both static and dynamic. The tomographic images are reconstructed using weight matrices, which depend on the source-detector geometry and flow velocity. These weight matrices are invariant to sensor geometry. Such approach may be successful for investigation and diagnosis of different multi-phase flows. This work is done in collaboration with the University of Bergen (Prof. G.A. Johansen).

Dr W. Mosorow has also developed, in collaboration with the Malaysian Institute for Nuclear Technology Research (Dr J. Abdullah, M.Sc. R. Mohd Zain), a novel approach to combine electrical capacitance and optical tomography to monitor and investigate solid/gas flow. The approach enables to gain a high quality image in full-scale concentration distribution of solid/gas flow. The algorithm is under development.

Electrical Capacitance Tomography is generally identified with conventional two-dimensional image of cross-section and with two-dimensional image reconstruction. The latest research has been focused on a new trend in modern tomography measurement systems which appear to be 3D tomography (see previous section). The reconstruction process is an extension of the methods described earlier in this section for 2D measurements. 3D-ECT is not a mature branch of process tomography yet, and the research carried out at the Institut by M.Sc. R. Banasiak and M.Sc. R. Wajman has considered the extension to 3D of forward problem solving.

Algorithm for Velocity measurements

Involvements: Prof. D. Sankowski, Prof. T. Dyakowski, Dr W. Mosorow, M.Sc. K. Grudzień.

The "best-correlated pixels" method is based on calculating the cross-correlation between a pixel from the first plane from twin-plane ECT system and some pixels from the second plane. The pixels from the second plane are chosen from the corresponding pixel and its neighbourhood satisfying the criteria max{RXn, Yn-j}j=B, where B is the neighbourhood of pixel n on the second plane. This simple method has a disadvantage. In practice, one can observe that some pixels from the first plane are best correlated with the same pixel from the second plane. In that case, these pixels are not used to calculate the velocity profile. An aleternative to this method is presented below.

The proposed approach is based on a model of flow wherein tomographic measurement is presented as stochastic processes. Using cross-correlation methods and "k-means" clustering algorithm based on ECT images (coming from twin-plane tomography system) extraction of the flow structure is allowed. Such structures can be observed as certain spaces between two planes of cross-sections called virtual channels. The flow structure detection allows the behaviour of the separated structures to be analysed as an alternative to the whole flow analysis. This concept can be used e.g. for investigation of turbulent flow propagation.

Data analysis of flow processes by ECT

Involvements: Prof. D. Sankowski, Prof. T. Dyakowski, M.Sc. K. Grudzień, M.Sc. A. Romanowski, M.Sc. Z. Chaniecki.

Dr Z. Chaniecki has carried out data analysis of flow process of friable materials in silo. The main results show that the type of a silo discharge process is a function of materials properties, initial solid concentration and size of the outer diameter and roughness of the silo wall. The latter parameter is a critical one affecting both the type of discharge flow and also the amplitude of generated mechanical vibrations in a silo wall. The data are also used to develop a physical model describing the changes in material concentrations for non-cohesive and cohesive particles.

In collaboration with the University of Leeds (Prof. R. Williams and Dr R. Aykroyd), M.Sc. K. Grudzień and M.Sc. A. Romanowski have investigated other type of powder flow - hopper flow. In this case they have investigated the behaviour of discharging of solid materials from hopper.

To investigate the powder behaviour as well as to cope with inverse problem they have firstly applied functional analysis (Euclidean Distance Matrix). Secondly, they have developed a new methodology based on Bayes' theorem and Markow Chain Monte Carlo methods. This approach, based on advanced statistical modelling such as Bayesian framework, is a powerful methodology and gives great flexibility in terms of physical phenomena modelling. Firstly it enables delivery of a practical stable solution to the ill-posed inverse problem with the aid of considerable prior information. Secondly it allows the estimation of the distribution of chosen parameters defining the characteristics of the industrial process without the phase of image reconstruction. This property has an important feature of making feasible implementation of automatic industrial process control systems based on process tomography.

This novel methodology was applied lately also to pneumatic conveying of powders based on experiments conducted at the University of Manchester.

Collaborations

Grant "Industrial Process Gamma Tomography"

The Institut is co-leading with International Institutes from USA, Norway, United Kingdom, Brazil, Malaysia and Mexico, a project funded by the IAEA (International Atomic Energy Agency). The project concerns the capacity to use dual-modality system composed of g-ray and electrical tomography systems to study industrial process of fluidized bed. The task of the institut is to develop software to analyse temporal tomographic images from this industrial process. The grant has been started in March 2003.

Grant "Spatial and Temporal Modelling for Electrical Capacitance Tomography"

M.Sc. A. Romanowski and M.Sc. K. Grudzień have been involved in a international project funded by the EPSRC (The Engineering and Physical Sciences Research Council). They have carried out research studies on the behaviour of discharging of solid materials from hopper (already described in the "Research programmes" section). This work was carried out at the University of Leeds and supervised by Prof. R. Williams and Dr R. Aykroyd.

Grant "Application of Tomographic Methods for Modelling Multi-phase Flows"

This project was funded by the Royal Society (2003-2005) and co-ordinated with the University of Manchester (UK). The grant contributed to fund short stays for both parties in the partner organisation, in order to share advances in the research project.

 

7th World Congress on Industrial Process Tomography in 2013 was held in Poland

7th World Congress on Industrial Process Tomography was succesfully organized by the Institut of Applied Computer Science. Congress took part in Cracow on 2-5 September 2013.

WCIPT7

return to the top


Editors:
Andrzej Romanowski
Krzysztof Grudzień
Radosław Wajman

Last modification:
2016-04-09 20:45:24, Radosław Wajman