Modeling and Fault Diagnosis for Pipelines and Networks


Cristina Verde


  • Ove Bratland, Singapore
  • Bruno Brunone, University of Perugia, Italy
  • Vicenç Puig, Universitat Politècnica de Catalunya, Barcelona, Spain
  • Cristina Verde, National Autonomous University Mexico, Mexico


This workshop is aimed at the interested audience in modeling and fault diagnosis in pipelines and networks, who requires a reference guide about the main concepts and standard approaches of model based and data driven fault diagnosis. Its goal is the introduction of recent theoretical and technological advances of pipeline integrity and leak localization together with challenges and opportunities in general that offer this application to young generations. The focus is the automation problems associated with the modeling, supervision, monitoring, and software development for pipelines and networks by considering the security and integrity of the installations as the main concern. In particular, the issues of modeling and advanced monitoring technologies in main lines, water demand networks and lines of multiphase flow are considered. Readers with some experience in the community involved in the field of fault diagnosis will also find it useful to learn some concepts and advanced tools recently developed and to create synergies to improve the integrity of pipelines and networks. Engineers and researchers looking for transferable fault diagnosis methods will also find the content of workshop valuable. The content of the workshop has been designed around real case studies of water and oil transportation lines and networks, and it is divided into four sections. To cover different points of view and information, four experts with diverse backgrounds will share their technical knowledge. The first one presented by Prof. Cristina Verde focuses on safe process fundamentals related to modeling, analysis and automatic fault monitoring in real time for general pipelines and networks. The rest of the sections cover three important applications.

  • Leak detection in main pipelines through the testing of the transient response, presented by Prof. Bruno Brunone.
  • Leak monitoring of water distribution networks through machine learning techniques, presented by Prof. Vicenç Puig.
  • Computational modeling and software development for pipelines monitoring, presented by Dr. Ove Bratland.


Despite the significant advances of supervisory control and data acquisition (SCADA) systems for fluid transportation by pipelines in the last decades, important issues associated with the modeling, instrumentation, automatic control and monitoring of pipelines and networks have been only partially solved. One of the causes is that a system behaves differently from the design stage mainly because of the presence of uncertain roughness, unbalanced measurements, and deviations in the density and viscosity of the fluid, as well as line age, lack of maintenance, and natural phenomena. Thus, governments, industries and society as a whole demand the design of advanced tools through hardware and software for the safe, automatic operation of pipelines and distribution networks. The automatic control community has responded to this demand. Thus, supervision systems and control for diverse pipeline scenarios, computational tools and devices have been introduced in IFAC conferencess and symposiums for the automatic monitoring and the correct predicting of the flow and pressure in pipeline and networks’ operation. In the 20th World IFAC Congress and the last two IFAC-SAFEPROCESS symposiums, specific sessions were dedicated to the issues associated with the modeling, security and monitoring of fluid transportation pipelines and networks. These contributions are the results of the effort and integration of different disciplines and research groups all over the world for improving the security of the installations. On the other hand, pipeline expansion and the market for pipeline integrity are projected to
grow at a compound annual growth rate of 4.3% in the next six years around the world. This includes oil, gas and offshore and onshore scenarios. This means new monitoring and control systems will be developed for considering actual wireless possibilities and composite materials for satisfying the safe demand of industry, government and society.

Introduction of flow behavior in pipeline as well as network and monitoring issues, Cristina Verde

This first chapter gives a small survey of methodologies, methods for fault detection and localization in pipelines and networks together with an overview of the topic. As a framework of the workshop, the physical model based on mass, moment and energy balance for one spatial dimension is introduced. Examples where the sensitivity of the leak location with respect to the head loss for diverse computational function are discussed. The main part of the chapter deals with the dominant, internally based systems starting from the first work published by Billmann and Isermann in 1984, and the importance of a quality modeling for fault diagnosis is discussed. Some examples based on the experimental tests in a hydraulic pilot pipeline are analysed and the applicability of the diagnosis systems based on the state of the flow and pressure at the lines’ extremes if the leaks are sequential is shown.

Transient test-based techniques (TTBTs) for fault detection in long transmission mains, Bruno Brunone.

This chapter will concern innovative technologies for detecting leaks and defects in long transmission mains. After a brief review of the existing techniques, attention will be focused on the TTBTs, which are based on the execution of ‘safe’ transients. It is worth of noting that ‘safe’ transient means the injection in the pipeline system of a small amplitude pressure wave (i.e., a pressure wave with an entity of a few meters of water column). TTBTs will be firstly examined by means of numerical tests, then the results of the validation in the laboratory will be shown. In the final part of the talk, the encouraging results obtained in live transmission mains will be discussed.

Leak monitoring in water distribution networks using machine learning techniques, Vicenç Puig

This talk will present different approaches to detect, estimate and locate water leaks (with the main focus on the localization problem) in water distribution networks (WDNs) using hydraulic models and machine learning techniques. First, the actual state of the art will be concisely revised. The theoretical basis of the machine learning techniques considered will then be introduced, as well as WDN hydraulic models usually used by WDN management companies. Different real WDNs and district metered areas DMAs will be used to illustrate the presented approaches using simulated and real scenarios. The talk will end with a discussion about the comparison of results of the different presented approaches and with the current and future research trends in the field of leak monitoring in WDNs.

Computational models and software development for dynamic system analysis, Ove Bratland

Some of the methods used for detecting leaks in pipelines and pipe networks rely on mathematical models for the fluid flow. Such models are widely used, but implementing them in useful computer programs is not without challenges. The main difficulties are explained below: For liquid flow: The friction, which plays an important role both in determining a pipeline’s capacity and in determining how potential leak-revealing disturbances propagate, can be quite inaccurate. This problem comes from the fact that friction is a function of what the pipe’s inner surface looks like in a wider sense, not only its so-called “relative roughness”, that is currently the most used parameter. Traditional friction calculations also ignore transient friction, and that typically leads to models underestimating signal damping. For gas flow: Nearly all commercial simulations of gas networks use steady-state models. Transient models also exist, but on close inspection, it turns out that the most common transient gas simulation models typically ignore part of the dynamics that would be useful in advanced leak detection software. For multiphase-flow: The last 30 years have seen some major improvements in our understanding of multiphase pipe flow, and this now allows us to send mixtures of gas, oil and water through quite long subsea flowlines. That has saved billions of dollars by reducing the number of or even eliminating permanent oil rigs in various offshore fields. The models have so far been less successful to design software-based leak detection systems. The presentation will show mathematical models for these three types of flow and discuss some of the challenges involved in utilizing them for various purposes.