Knowl Inf Syst (2007) 12(2):255–278 DOI 10.1007/s10115-007-0076-4 Knowledge and Information Systems R E G U L A R PA P E R Matı́as Alvarado · Miguel A. Rodrı́guez-Toral · Armando Rosas · Sergio Ayala Decision-making on pipe stress analysis enabled by knowledge-based systems Received: 7 February 2006 / Revised: 9 November 2006 / Accepted: 25 November 2006 Published online: 21 March 2007 C Springer-Verlag London Limited 2007 Abstract This paper presents engineering decision-making on pipe stress analysis through the application of knowledge-based systems (KBS). Stress analysis, as part of the design and analysis of process pipe networks, serves to identify whether a given pipe arrangement can cope with weight, thermal, and pressure stress at safe operation levels. An iterative process of design and analysis cycle is done routinely by engineers while analyzing the existing networks or while designing the process pipe networks. In our proposal, the KBS establishes a bidirectional communication with the current engineering software for pipe stress analysis, so that the user benefits from this integration. The stress analysis knowledge base is constructed by registering the senior engineers’ know-how. The engineers’ overall strategy to follow up during the pipe stress analysis, to some extent contained by the KBS, is presented. Advantages in saving engineering man-hours and usefulness in guiding experts in pipe stress analysis are the major services for the process industry. Keywords Decision making · Pipe stress analysis · Knowledge-based systems 1 Introduction Engineering practice in the oil and gas industry is a combination of complex and specialized areas. Nowadays, the use of engineering software by engineers carrying out information input or manipulating decision variables is a daily issue. When M. Alvarado (B) Centre of Research and Advanced Studies (CINVESTAV-IPN), Av. Instituto Politécnico Nacional, 2508 Col. San Pedro Zacatenco, 07360 Mexico City, Mexico E-mail: matias@cs.cinvestav.mx M. A. Rodrı́guez-Toral · A. Rosas · S. Ayala Instituto Mexicano del Petroleo, Mexico City, Mexico 256 M. Alvarado et al. an engineering problem requires a specific solution, human engineers manipulate and interpret the input data to a software system and they decide to accept data for implanted solutions. Whenever the engineering software does not come up with an acceptable solution, engineers manipulate variables and parameters to achieve a convenient output that fulfils the mean requirements. The software tools, simulators, largely act like black boxes answering yes or no based on the input data with respect to some goal solution. However, no suggestion is delivered from the simulators about what to do if the input data does not assure a required solution. Actually, the relevant human expert task is dealing with the change of input data, typically during an iterative process of successive approximations to solution, so that new data gets closer to obtaining the required solution. 1.1 Interaction between human expert/engineering software This interaction process, technical expert–engineering software–technical expert, is common in engineering practice. In engineering companies—as in diverse productive organizations—it is well known that the human technical skills are finite resources that should be well managed in order to optimize the company work execution. In this paper, we show that the usage of modern artificial intelligence (AI) techniques rooted in KBS for specific engineering discipline is a key aspect in boosting productivity. AI soft-computing techniques like genetic algorithms, neural networks, and fuzzy logics do flexible data manipulation and support robust solutions in complex domains like those from engineering fields. Besides, the preservation of the engineering technical knowledge and improvement of the effective usage of preserved knowledge is strategic to promote sustainability of engineering services. Based on the knowledge management embedment in engineering practice, reuse of the human experts’ know-how for future works can be done in an effective manner. In this proposal, the effective use of a KBS is with respect to two main objectives: • To support a successful transfer of knowledge from senior expert engineering designers to junior designers. • To allow the human expert to use knowledge skills and experience from others within the organization without sacrificing the productivity and design quality. Junior designers can use a KBS to efficiently do their job without having to pass through a mean steep learning curve. Furthermore, in a broad perspective, KBS can be a powerful and versatile tool for the designer; the KBS allows him or her to reuse the know-how from other organization members, thus preserving design quality and augmenting productivity with the (semi-) automated support from the KBS. A point of view about the so-called expert system (ES) as the former generation of current KBS is that ES was used to automate the usage of information for low-level expert’s decision-making. Interest in old ES is renewed by the current challenge of dealing with complex dynamics in business process re-engineering as well as in concurrent engineering in an automated fashion. Due to the maturity of the AI methods, practical aspects relevant to engineering can be addressed jointly by software developers, knowledge engineers, and domain experts throughout KBS deployment. Decision-making on pipe stress analysis enabled by knowledge-based systems 257 1.2 Pipe network: critical lines, flexibility, and supports distribution Opportunities were identified for KBS application in pipe stress analysis (PSA) which is a part of the design process of pipe networks. PSA is used to identify whether a pipe arrangement will cope with weight, thermal, and pressure stress at acceptable levels under engineering design standards for safe operation. An iterative process of design and analysis cycle is done on the existing or to-be designed pipe networks. The KBS can support engineers during the steps of process pipe design by suggesting pipe design rules or possible solutions, such that design conditions for safe and reliable operation will be accomplished in an effective manner. The more the KBS guides the verification and approval of design proposals, the more the pipe stress analysis expert’s time is saved. In addition, KBS can support PSA experts in routine or preliminary tests of pipe network’s design reliability. The PSA concerned in this paper, the artefact to be analyzed, is the design of process plant pipe networks in the upstream or downstream petroleum industry. According to our PSA experience, decision-making bears on • • • • identification of critical pipelines, top-down ranking of critical pipelines, flexibility of the piping system for relaxing overstressed points, proper and balanced supports distribution. The PSA engineers’ decisions are founded on the assessment and interpretation of design information of the piping system. Such information is provided from the engineering workgroups at previous steps of the pipe system design. Whenever a formal PSA of the pipe design indicates an overstressed point, decision-making over possible solutions imply modifications of the pipe pathway, supports distribution or both. If none of these modifications solve the pipe overstress, a redesign coming from the steps of basic engineering would be necessary. The KBS has a bidirectional communication with the PSA software simulator. Through the rest of the paper, the overall strategy followed during the development of the KBS for PSA and the advantages in its application are presented. Beyond saving PSA engineering man-hours, KBS increases productivity and becomes useful in the process of training new technical experts. As an antecedent of our aims, we mention that KBS technology has successfully provided intelligent support to humans during the process of database analysis and design [36]. However, skepticism remained on the capacity to simulate the diagnostic competence of human designers. Expert human designers employ the so-called knowledge of the real world in carrying their design activities. Last advances in AI on knowledge management and knowledge representation techniques in modern KBS apply on implementing such capacity. Herein this advance is advantaged by linking the capabilities of highly experienced people on pipe stress analysis with those from AI. This has methodological advantages including project development management, human expert use, and acceptance (most likely, since they are contributors too), and practical as well as realistic user requirements for the KBS. In the next section, a summary of the state-of-the-art KBS in engineering for the process industry as well as the presentation of the AI techniques for decision-making is presented. Throughout the rest of the sections, our pipe stress analysis decision-making perspective is introduced. 258 M. Alvarado et al. 2 Engineering decision-making After mid-1970s, interest in ES decreased; they recovered some popularity in the 1990s with the application of techniques like case-based reasoning and constraint-based reasoning. These were mainly associated with business process re-engineering and knowledge-based software engineering. Expert systems, now recognized as KBS, can capture the (senior) expert’s know-how on design solutions; they involve the construction and usage of the following modules: • Knowledge base (KB), • Inference engine, • User’s interface. An engineering KB contains basic as well as complex facts forming the knowledge about the engineering domain, e.g., chemical, mechanical, civil, and so on— at the basic or detailed levels of engineering. The inference engine codifies the way the expert engineers do reasoning in such a way that from the set of KB information, new relevant information can be induced, deduced or abducted. The knowledge and reasoning process is the qualitative part dealing with semantic objects that require definition or translation to numerical values. In addition, the KBS should perform some of its inference based upon dynamic changes in knowledge; the execution order of the system should not be defined solely by data, as with many algorithms. The user interface sets the way the input information must be captured for material problem situations. 2.1 KBS for process industry KBS on engineering applications for the process industry herein is reviewed, specifically in relation to piping systems, process plants, and equipment. Applications include the analysis of metallurgical failures in an ES for recognizing modes of failure like stress corrosion and hydrogen embrittlement [31]. Later, the development of an ES designed to assist technical personnel in the evaluation of the physical integrity of process equipment by generating diagnosis and explanatory information was reported in [17]. In this ES, the knowledge base draws in the three expertise domains required to evaluate the integrity of the equipment: inspection of the equipment, numerical analysis of critical defects, and recommendation of corrective actions. A KBS for material selection in an engineering design process is described in [44]. It discusses the development of material databases to be used as material selection packages. Examples were shown for the use of a KBS in material selection in the domain of polymeric-based composite. The importance of KBS in the context of concurrent engineering is also explained. 2.1.1 Computer aided design tools The concept of so-called intelligent computer-aided design (CAD) systems has been identified as an approach toward integrated engineering environments. The experience of designers is the main tool in the process of finding an optimum Decision-making on pipe stress analysis enabled by knowledge-based systems 259 route of ship pipes, which is a complicated and time-consuming process. To reduce the amount of design man-hours and human errors, an ES shell and a geometric modeling kernel were integrated for design process automation [23]. They implemented methods of ES to find the routes of ship pipes on the main deck of a bulk carrier. A framework of the intelligent CAD system for pipe auto-routing was suggested. The CADDS 5 of Computervision is used as the overall CAD environment, the Nexpert Object of Neuron Data is used as the ES shell, and the CADDS 5 ISSM is used to build user interface through which geometric models of pipes are created and modified. 2.1.2 Knowledge-based engineering Competitive pressures are forcing equipment manufacturers to reduce product development times, minimize design iterations, and react rapidly to changing markets. Concurrent engineering replaces the traditional sequential design process with parallel efforts in multiple disciplines, increasing the product quality while reducing the work time. Knowledge-based engineering captures product and process knowledge contained in the ‘corporate memory’ to enhance and accelerate the design process. Linking these two together provides a wide variety of synergistic effects. A general description of the process used to create a knowledgebased engineering (KBE) for concurrent engineering (CE) is given in [27]. The use of the system to solve real world design problems in compressor rotor design is discussed. Selecting a shell-and-tube heat exchanger type and geometry is an application, where a wide range of specialized knowledge is available as qualitative rules that can be incorporated in an ES. Abou-Ali and Beltagui [1] present a technique for building a KBS utilizing object-oriented ES shells. The constructed prototype assists the designer in making decisions about fluid allocation, selection of Tubular Exchanger Manufacturers Association (TEMA) shell type, bundle, heads, and various geometrical details. The final aim of developing an ES of this type is to achieve an integrated design procedure, from initial selection to the final thermohydraulic and mechanical design. Kim et al. reported an ES called NPiES for nuclear piping integrity. In this work, the structure and development strategy of the ES including, user interface, a database (where nuclear piping material properties are stored and the unknown material properties are provided through inferring the known material properties), a knowledge base (with rules for inferring material properties), expert part (where the most appropriate evaluation method for given input condition is recommended), and finally, an integrity part (where they plan to do the evaluation of piping integrity), is described. 2.1.3 Case-based reasoning for experience recording Case-based reasoning (CBR) employs past problem-solving experiences when solving novel problems. In [40], CBR has been applied to mechanical bearing design. Their system retrieves previous design cases from a case repository and uses adaptation techniques to modify them for satisfaction of problem requirements. The approach combines (a) parametric adaptation, to consider parameter 260 M. Alvarado et al. substitution and the interrelationships between the problem definition and its solution; (b) constraint satisfaction, to globally check the design requirements to assess case adaptability. The system was implemented and tested in the domain of rolling bearings. The application of KBS to the task of failure analysis and design against failure is reviewed in [20]; they emphasized on the reasoning methodologies and the knowledge domains. Case-based reasoning techniques were considered to be the most suitable for generic failure analysis due to the complexity of knowledge required. They concluded that future trends in diagnostic expert systems will be based on the holistic hybrid rule-case-based reasoning approach combined with a number of stand-alone engineering failure-analysis calculating software tools and a multimedia-type KBS for different failure modes. 2.1.4 Knowledge management practice Knowledge management (KM) grants effective methods to organize and reuse information for executing business procedures [33]. In the process industry, the complex interrelationship between design stress and environment often makes this an overwhelming task, even for engineers with considerable know-how and experience in failure analysis. Usage of information systems as complementary frameworks for the knowledge-based ones, facilitates the analysis task. The stored and subsequently accruing expert’s experience is available to be reused for junior engineers. Furthermore, organized information can be used for planning, decisionmaking, and process optimization. As an instance, the building of a KM system in the Mechanical and Industrial Department at DAR AL HANDASAH, a leading consulting firm in the Middle East, as well as lessons learned in building the system and the steps needed to improve it are reported in [29]. 2.2 Decision-making representation 2.2.1 Artificial intelligence techniques Nowadays, the codification of autonomous agent’s knowledge as well as the reasoning capacities is strongly based in AI techniques like multivalued logics, genetic algorithms and neural networks. The autonomous agent system’s application in any discipline carries on the usage of AI methods to deal with diverse information in a flexible manner. In engineering, the expectation is the advancement in the knowledge-based engineering added value by applying tools that are integrated in autonomous computational modules. Currently, in engineering, the use of simulators is ubiquitous making easier the test of designed solutions. The input data configuring a proposed solution are processed in the simulator for validation if conditions for safe operation are fulfilled. Simulators process huge volumes of data that typically implies hard and complicated calculus, in a precise manner, making it easy—to some extend—for the human analysis of the solutions. However, there is no commercial system guiding the assessment of alternative solutions, either at the preliminary tests or during the iterative steps looking for Decision-making on pipe stress analysis enabled by knowledge-based systems 261 a convenient solution. Such guiding system is thought to suggest possible adjustment of parameters in order to fit the design of an intended solution. In pipe stress analysis, for instance, the designer has to decide about the pipe flexibility as well as the distribution of balanced pipe supports. The intensive usage of AI methods supporting the interpretation and assessment of technical and administrative data is required, such that the data becomes useful information through the analysis of novel solutions for either well-known or novel problems. Furthermore, a step forward in this direction is to be able to codify in computational systems, the experiences of engineers participating in routine operations, but specially, the acquired know-how of engineers providing novel solutions to challenging problems in their expertise area. 2.2.2 Architecture Decisions in engineering are related to an artefact being designed, constructed, operated, or maintained [5]. Therefore, there should be a link that exposes the decisions made, but explicitly with respect to the issues in artefact decision-making. Systems like QOC [28], KBDS [6], n-DIM [47], DraMa [12], and CLiPS [7] have addressed this relationship and have proposed different solutions. A relevant conclusion is that the relationship between the designed device and the decisionmaking about such a design must be a part of the representation. Moreover, in [2], is considered the automation of a framework to embody decision representation by integrating 1. technical information about device and issues to attend, 2. workflow diagram, 3. the knowledge that is being shared between members of the organization. According to such architecture, in our approach to pipe stress analysis decision-making, the software simulator contains technical information about the piping system (1) to be analyzed with respect to (2) static and dynamic effects, whereas (3) the human expert’s cognitive abilities are coded and then applied throughout the KBS (see Fig. 1). Furthermore, execution of tasks throughout the workflow carries on decision-making processes at different levels of complexity. In turn, the decision-making process is largely based on the participants’ mutual knowledge. The base of architecture combines the widely accepted state task network (STN) representation that associates one or more issues to each transition in the network [4]. Every issue has a number of possible solutions (options), and a decision involves the selection of one of these options. A decision is implemented throughout an action that is the embodiment of the transition between states. Thus, decisions/actions are the transitions linking the issues/states. In addition, the record of the decisions and their components (e.g., the options considered and the criteria used for their evaluation), also known as decision rationale, could be the most convenient way to record best practices and lessons learned. Decision rationale would be an essential part of knowledge management, the key principle of which is capturing intellectual assets for the tangible benefit of the organization. The suitability of KM based on decision-rationale to improve the 262 M. Alvarado et al. Fig. 1 PSA decision-making by KBS competitive edge of engineering is allowed so that the aim of any design engineering organization may produce projects with high quality and make them available in less time. 3 Pipe stress analysis PSA is a complex engineering discipline which covers the design, analysis, and identification of piping problems by ensuring that weight, thermal, and pressure stresses are at acceptable levels specified in engineering design standards. PSA includes the calculation of piping code stresses, loads, and deflections under static and dynamic loading conditions. The stress analysis of pipe networks is normally done using the finite element method (FEM) [35]. The reasons for the analysis of Pipe stress on a piping system is essential to ensure that the piping is well supported and does not fall or deflect under its own weight; the deflections are well controlled when thermal and other loads are applied; the loads and moments imposed on machinery and vessels by the thermal growth of the attached piping are not excessive; and that the stresses in the pipework in cold and hot conditions are under the range allowed. PSA addresses problems such as thermal analysis (analysis for free and restrained thermal growth conditions); deadweight analysis (analysis at ambient temperature with a system of hangers at specific locations to support the weight of the system, for allowable stress and reactions at equipment connections); seismic analysis (static or dynamic); wind load analysis (static stress analysis); transient analysis (for various transient loading conditions such as, turbine trip, pipe whip, safety relief valve trip, etc.). Static analysis in PSA includes the use of hangers, wind load sets, nozzle flexibilities and stresses, equipment load check (under engineering standards, for example, for steam turbines (NEMA SM23), centrifugal compressors (API 617), air-cooled exchangers (API 661), etc.), flange leakage and stresses, fatigue analy- Decision-making on pipe stress analysis enabled by knowledge-based systems 263 sis and cumulative usage (to calculate the remaining life based on material fatigue curve data and an assigned number of cycles), offshore piping analysis (for analyzing individual pipe elements experiencing loading due to hydrodynamic effects of ocean waves and ocean currents). Dynamic analysis in PSA considers dynamic data such as lumped masses, imposed vibration, snubbers, and spectrum definitions. Dynamic analysis includes aspects such as mode shape and natural frequency calculations (for reviewing the systems natural modes of vibration), harmonic forces and displacements (to evaluate the vibration response of a damped system to a range of harmonic forces or displacements to simulate mechanical and acoustic line vibrations), shock spectrum analysis and independent support motion (including anchor movements), force spectrum analysis (for the analysis of general impact loads such as water and steam hammer, slug flow and relief valve discharge), modal time history analysis, relief valve load synthesis (to calculate the dynamic thrust load and transient pressures from relief valves in open discharge systems). Engineers should also combine different static/dynamic loads in order to properly address the occasional load requirements of the piping codes. 3.1 PSA: from early to current computer applications Early applications of PSA on microcomputers are reported in [3]. In 1955, the stress concept for evaluating thermal expansion stress was recognized by an international engineering code, for pressure piping [38]. Applications of pipe stress analysis include pipelines with soil forces and longitudinal/lateral pipe movement [37], pipe stress/support analysis to establish extra safety margin [26], and underground pipes in granular or sandy soil using a pipe stress program for the evaluation of thermal and pressure effects [42]. The idea of a fully integrated engineering software has been reported [13] on the AUTO-PIPE CAE System that allows the user to perform the entire sequence of piping analysis and design in a streamlined work flow process. Tasks in this automatic process include pipe stress analysis, pipe support location optimization, stress isometric drawing generation, pipe support pattern selection and member design, 3-D interference detection for support. At the core of the system is the AUTO-PIPE (relational) database which contains all the static (project-specific) and dynamic (model-specific) data required for all of the mentioned tasks. The AUTO-PIPE CAE System has been used for pipe system design of nuclear power plants in Japan to achieve substantial manpower reduction and cost savings. Now, there is a commercial software, AutoPIPE, as a stand-alone computer-aided engineering (CAE) program for the calculation of piping stresses, flange analysis, pipe support design, and equipment nozzle loading analysis under static and dynamic loading conditions. In [45], stresses of a pipe flange connection with a spiral-wound gasket under internal pressure were analyzed. It acknowledged the nonlinearity and hysteresis of the gasket by using an axisymmetric theory of elasticity and the FEM. Knowledge advances are still active in this area; for example, when branch connections in low-pressure large-diameter piping systems are designed, as reported in [43], the flexibility factors in ASME B31.31 for branch connections do 264 M. Alvarado et al. not assist the designer in taking credit for flexibility that may exist in a large diameter intersection. The author reports that since the stress intensification factors (SIFs) are relatively high for large-diameter piping, many stub-in branch connections will require a pad to meet the code displacement stress limits. 3.2 Engineering software Pipe stress analysis can be done using analysis software such as AutoPIPE or CAESAR II. The model is constructed from piping general arrangement, piping isometric drawings, and piping and valve specifications. Once the system is modeled and the boundary conditions are set, comprehensive stress analysis calculations are done by the engineering software, and modifications to the model can be made to ensure compliance with the design requirements. Many engineering and energy organizations, around the world engaged in services on design and analysis of process pipe networks use the CAESAR II engineering software, first introduced in 1984 by a company named COADE; today, it is perhaps the most used in the engineering area [14]. CAESAR II allows the analysis of piping systems subject to weight, pressure, thermal, seismic, and other static and dynamic loads. The code compliance report generated by CAESAR II defines the overstressed points in the system. CAESAR II begins a static analysis by recommending load cases necessary to comply with piping code stress requirements. As a comprehensive program for pipe stress analysis, it includes a full range of the latest international piping codes. It provides static and dynamic analysis of pipe and piping systems, and evaluates fiber-reinforced plastic (FRP); buried piping; wind, wave, and earthquake loading; expansion joints, valves, flanges, and vessel nozzles; pipe components; and nozzle flexibilities. The program automatically models structural steel and buried pipe, and provides spectrum, time history analysis, and automatic spring sizing. CAESAR II includes component databases and an extensive material database with allowable stress data. It also includes a bidirectional link to COADE’s CADWorx Plant drafting package. The program’s interactive capabilities permit easy evaluation of input and output, a valuable match for the iterative ‘design and analysis’ cycle, and has an easyto-use menu-driven interface. Context-sensitivity helps provide instant technical assistance. Data values depicted in the help screens are automatically presented in the current set of units to make input easier. 3.2.1 About exchanging data from CAESAR II CAESAR II offers a link to CADWorx/PIPE, COADE’s AutoCAD based pipe drafting and design software. This is a fully functional, bidirectional link between CAD and the PSA program. CAESAR II has a neutral data file format for independent use in exchanging data with other programs such as CadCentres PDMS and Jacobus 3DM. Piping input and output can be directed to an ODBC database, R e.g., MS Access , for data review and manipulation outside CAESAR II. Decision-making on pipe stress analysis enabled by knowledge-based systems 265 3.3 Expert systems in pipe stress analysis and design The identification of piping design rules and how these rules can be incorporated into an expert system using a common subset of LISP was reported for an expert system that was then interfaced with a computer-aided design package [21]. Also, applications in engineering companies like Brown and Root, Inc., U.S.A., reported the opportunity for AI techniques in diagnosing high-energy piping problems [41], arguing the need of an expert system for efficiently using their company’s experience on many high-energy piping systems in fossil power plants, technical papers, procedures, reports and reviews with a large database, including the accumulated experience of senior engineering specialists. The challenge was how to best use this expanding base of valuable knowledge and experience. They saw an opportunity for piping data and expertise to be used more efficiently, comprehensively, and accurately with an expert system. An expert system environment designed to integrate multiple sources of knowledge required to analyze the internal structure of flexible pipes named FRAES is reported in [9]. There, numerical algorithms, databases, and expert knowledge are explored by the inference mechanism of the system to assist the technical personnel of petroleum companies in the analysis, design, and diagnosis of flexible pipes; these are used as flowlines or risers in offshore applications. Diab and Morand (2005) proved that a safety factor principle is enough to analyze safety reserves in buried pipes because of the variation of the phenomena acting on the behavior of the pipe sewers. Their decision support system is used to boost the efficient use of existing resources as it integrates all of the information involved in a decision-making process. They report a semiprobabilistic approach to diagnose urban sewers, which is divided into two parts: one based on a simplified probabilistic method (concerns only the mechanical behavior of the pipe); the other part is based on the established rules to integrate the impact of the pipe behavior on its environment. They insist that their method will permit to establish a rational diagnosis of urban sewers. Actual decisions that address the human expert and the expert systems for pipe networks and pipelines design include a range of complex and specialized knowledge like the one outlined in the next section, thus showing a necessity for the development of a KBS as proposed here. 3.4 Issues and options on piping system design As in many engineering disciplines today, the expert uses computer software for engineering calculations, then he or she may need to decide on the modifications required to apply to the computer model until a satisfactory solution comes up. Technical expert in PSA should know when and how to use specific restraints (or support types) for piping systems and these include the following. Restraints: A device that prevents, resists, or limits the free thermal movement of the pipe. Restraints can be either directional, rotational, or a combination of both. 266 M. Alvarado et al. Anchors: A restraint that provides substantial rigid strength, ideally allowing neither movements nor bending moments. There are also anchors with displacements. Expansion loops: A purpose designed device that absorbs thermal growth; usually used in combination with restraints and cold pulls. Cold pull or cold spring: It is used to pre-load the piping system in cold condition in the opposite direction of the expansion, so that the effects of expansion are reduced. Cold pull is usually 50% of the expansion of the pipe run under consideration. Cold pull has no effect on the code stress but can be used to reduce the nozzle loads on machinery or vessels. Spring hangers: They are used to support a piping system that is subjected to vertical thermal movements. Variable effort spring hangers are usually incorporated for vertical thermal movements up to approximately 50 mm. The variation between the preset and operating loads should not be more than 25% of the operating load. Constant effort spring hangers are usually incorporated for vertical thermal movements in exceeding 50 mm. Solid vertical support: It is used in places where vertical thermal movement does not create undesirable effects or where vertical movement is intentionally prevented or directed. Human expert should also know aspects of solid supports in the form of rods or pipe shoes, the importance of free horizontal movement of the pipe as not being impeded unless the horizontal restraint is desired, harmonic forces and displacements influencing the vibration response of a damped system, etc. The expert should be able to transfer his design to/from hydraulic analysis department (using commercial software too, e.g., Stoners LIQT and Sunrise Systems PIPENET or even by hand, expressing the piping isometric into his own engineering software, e.g., CAESAR II. He or she should know specific topics from engineering codes, as well as effects like single or double acting translational, single or double acting rotational, translational with bilinear stiffness, use of snubbers (shock absorbers), guides and limit stops, bottomed-out springs, tie rod assemblies, gaps and friction, connecting nodes for nodal interdependence and large rotation rod supports. During the piping design stage, a choice of (algebraic) combination of displacements, forces, and stresses results in the modification of load cases. Some choices are indicated as obliged, but there are others that admit alternative solutions, and the criteria to select them is a key judgment by senior engineers. On the other hand, to pipe the support’s distribution should be decided on the position of each support, the type of support, the distribution along the process plant, etc. 4 The knowledge-based system The PSA experts’ decision-making concerns with the artefact to be analyzed, which is the design of process plant pipe. The steps to PSA decision-making are the ones mentioned in Sect. 1.2: • identification of pipe critical lines, • ranking of critical lines according to the level of risk or danger, Decision-making on pipe stress analysis enabled by knowledge-based systems 267 • flexibility of the piping system as for not having overstressed points, • proper and balanced supports distribution. PSA engineers assess and interpret the information on the design of the piping system delivered from the basic engineering (process and design) technicians. PSA engineers do decide whether according to the given specifications, the pipe pathway is reliable or not, as well as if the distribution of pipe supports is well suited, then will that pipe operate safe. Identification of pipe critical lines indicates the part of the pipe to put special attention considering the top-down ranking of risk and danger of the pipelines. Whenever PSA of the proposed design derives in pipe overstress, the decision-making compels to modify pipe pathway or supports distribution or both, as the solution that the PSA engineers can introduce to achieve safe operation. 4.1 Device information and experts know-how Technical data about the piping system to make decisions is broadly contained in the PSA simulator jointly with other engineering software tools. PSA simulator uses the design information of the piping system to test if it suits stress enough below the allowed limits. If design data input to simulator fulfils the specified restraints and there are no overstressed points, then the simulators output is OK. Otherwise, reasons for a negative answer are not shown. However, no recommendation is indicated about possible changes to introduce in the proposed pipe design. Actually, this is the current difficulty that PSA engineers deal with. Currently, assessment and interpretation about what to do is being obtained from human experts’ know-how. The more the engineer’s experience, the more quickly the required design is found out: junior engineers can spend a lot of man-hours to get the right solution, usually by a trial–error cycle or by asking senior PSA engineers for some guidance. Alternatively, we experienced that large part of the routine and/or fine decision-making can reside in the KBS. When the PSA software simulator finds overstressed points in the piping system, the human expert feeds the simulator with alternative data. The experts’ recommendations, besides the processed information to find them, can be coded inside the KBS for PSA. Like the human experts, KBS will support decision-making; thus, it should deal with the assessment and interpretation of information on the design of the piping system to address an enough-flexible pathway as well as a proper distribution of supports as human engineers deal with. As a human supporter, KBS should guide, to some extent, the eventual changes that could be introduced in the pipe design. KBS takes as input data the simulator output, and fashions possible changes to avoid overstress; then, this new data of design is the next input to simulator in turns. This way, an interacting cycle—simulator/KBS/simulator—receives after some iteration, a well-suited pipe design. Symmetrically, the initial data that feeds the simulator can be previously assessed by a KBS so that a KBS/simulator/KBS cycle works as well. Then, KBS offers a possible design of the solution as the input for the simulator, or that the simulator output is the input to be assessed and weighted for the KBS. 268 M. Alvarado et al. Fig. 2 Developers of the pipe stress analysis KBS 4.2 Pipe stress analysis reasoning Based on the concepts of [16], people involved in the development of KBS for pipe stress analysis are (a) senior experts with engineering experience, (b) knowledge engineers, and (c) the end users. In all the cases, more than one person would participate, since the complexity and magnitude of each PSA matter is wide and complex enough (see Fig. 2). The knowledge acquisition tool (KAT) is for KB construction, and the KAT shell serves the purpose of constructed knowledge base [46]. Backend of the KAT is based on fuzzy sets and logic that provides a powerful support to KBS inference engine. Because of fuzzy sets, the parameters used to model or simulate an engineering situation can have an ad hoc range of values. Fuzzy logic furnishes the parameter’s combination in such a way that a global assessment of the engineering problem is available. There, a sample of rules that human expert uses for decision-making in designing pipe networks is presented. They are being implemented in the knowledge base and grouped in the steps mentioned in this section. 4.2.1 Critical lines In the very first step are identified the parts of the pipe that, due to the flow conditions, material, and size of the pipe as well as the type of connected devices, is especially dangerous and needs extra care. Flow conditions refer to temperature, pressure, type, toxicity, density, regime, among others. Pipe size involves diam- Decision-making on pipe stress analysis enabled by knowledge-based systems 269 Table 1 Rules to identify critical lines Rule 1. On the selection of critical line subject to PSA because high flow pressure and temperature. IF pressure is >15 kg/cm2 man. OR operating temperature is > 150 ◦ C.OR. below −10 ◦ C .THEN. critical line . AND. do PSA Rule 2. On the selection of critical line subject to PSA due to big diameter size. IF pipe line diameter >20 in .THEN. critical line .AND. do PSA Rule 3. On the selection of critical line subject to PSA based on pipe material of construction. IF pipe material of construction is different from carbon steel .THEN. critical line .AND. do PSA eter and length. Connected equipments to the pipe are furnaces, bombs, thermal changers, turbines, and compresors, among others. Some example rules for this step decisions are presented in Table 1. 4.2.2 Top-down ranking of critical lines The second step is the creation of a top-down scoring of lines’ criticalness level due to the characteristics of the flow each line transports as well as the pipe diameter or the equipment that is being connected. Most combinations making pipe critical lines involve high flow temperature, medium or big pipe diameter as well as more fine equipment being connected. The more the level of each flow temperature, pipe diameter size or equipment fineness, the more the level of line criticalness. By the process of Cartesian coordinates, Fig. 3 illustrates that point A corresponds to a more critical line than the one represented by point B because A contains flow with a higher temperature, has a bigger diameter, and connects more fine equipment than does B. Also with equal temperature, point D sets a more critical line than does E because D’s diameter is bigger. As a recommended practice, the most critical lines must be first OK designed and then constructed. This way, the pipe pathway space needed to locate critical lines or the facilities construction required to keep them together with all compul- Fig. 3 Combination provoking critical lines 270 M. Alvarado et al. sory conditions can be allowed without restrictions. As much as most critical lines are designed or constructed, the lines of next minor level of criticalness should be designed; the design of less critical lines can be adapted to the left conditions after the design and construction of the most critical ones. Experiences doing the design and some times the construction—regardless of this order—advice that hard difficulties in constructing extreme critical lines may occur. 4.2.3 Pathway modification Modifications about pathway aim to benefit pipe flexibility by introducing pipe architectural elements like the following among others: • expansion joints, • expansion loops, • thermal changers. Typically, a long straight section in pipe is not recommended; it facilitates the increment of flow pressure or force, thus increasing pipe stress. In this case, it is suggested to modify the long section architecture by introducing an expansion loop or omega (for the letter shape). The loop diminishes the flow inertia and reduces the pressure or temperature. An alternative is the introduction in the middle of the pipe long section of an expansion joint made of a more flexible material with shape of folds or biome that augments the pipe flexibility. Actually, both of the mentioned resources augment the pipe flexibility in order to cope with the stress flow induced by pressure or temperature. Also it guarantees a safe pipe operation in front of the flow turbulence or the so-called flow ram touch. Other usual pipe circumstances concern with the flow regime, namely liquid, gas, or even solid. Connection to a specific equipment to modify other flow regime is required. In these cases, it is a condition to attend vendor’s specifications about the devices to guarantee a right usage. Other aspect to be attended is the corrosion that pipe is exposed to. Special materials covering the pipe’s inner surface should be considered such that enough pipe resilience is assured. 4.2.4 Supports modifications Balance and equilibrium on pipe’s weight and stress also concerns the supports pipes keep. Right distribution of supports as well as the adequate support at the required place contribute to a safer pipe operation. Actions that PSA engineers can take with regard to pipe supports in order to achieve safe operation as follows: • add supports, • change supports separation or distribution, • change the type of support. By adding supports or changing the separation between them, a well-suited weight distribution can be obtained. On the other hand, when the pipe height is a variable, a below-fix-pipe support is not suggested but a pipe above the flexible one is. Pipe section’s height varies due to pipe expansion or distension from variable temperature from the inside flow or environment. This is a typical situation of weather conditions, like the ones of desert, where severe changes occur from day Decision-making on pipe stress analysis enabled by knowledge-based systems 271 Fig. 4 Overhead piping system for a distillation tower to night; variation in temperature is high during a 24-h cycle such that special supports adaptable to different conditions are needed. Example. One of the first process operations in a petroleum refinery is performed in an atmospheric distillation column, whose separated vapour goes to a condenser through a piping system as the one shown in Fig. 4. The piping system consists of one feed and one discharge connection pipe segments, one rigid Y support, three shoe supports, and four long-radius elbows. The upper end is connected to the top of the distillation column while the piping system’s lower end is connected to the overhead condenser. The resultant pipe stress analysis done with CAESAR II determined that forces in the piping system exceed the limit specified in the ASME B31.3 code for process piping. The pipe system does not have sufficient flexibility to accommodate the elongation of the atmospheric tower resulting from the temperature variations. The engineering expert solution incorporates a variable spring hanger to permit upward movement caused by the elongation of the tower; then, CAESAR II resulted in no warning or error messages for exceeding code limits. An alternative solution was the introduction of an expansion joint, it was an expensive solution— therefore not practiced—due to the pipe diameter. Nor was an expansion loop introduced, because it would be necessary as an additional support to a great height, which is expensive too. 4.3 Major redesign Whenever none of the PSA proposals solves the pipe overstress, a redesign must be practiced by the process and/or design engineers. It passes on major modifications of the pipe pathway concerning such steps of the pipe deployment. 4.4 Intelligent Chat KBS/PSA As introduced in Sect. 2.2, the integration of technical information about the designing issues, in this case, the process workflow and the experts’ know- 272 M. Alvarado et al. Fig. 5 KBS/PSA chatting how, harmonize the holistic solutions that benefits decision-making for our purposes. As mentioned in Sect. 4.1, the technical information on device under consideration is modeled in the software engineering PSA simulator. On the other hand, the expert engineers’ information is partially included in the KBS as long as it captures the human expert’s know-how. Rules to identify critical lines as well as for the top-down scoring fashion or the ones about the pipe pathway modifications and support distribution codify the PSA engineers’ know-how. This way automation of the interaction between the parts of the proposed architecture is arranged. The managed knowledge about pipe stress analysis is coupled with the pipe network being designed. The KBS becomes a smart mediator between the simulators modeling and the PSA engineer experiences adjusting the model (see Fig. 5). KBS can suggest alternatives in order to achieve an adequate piping system design under the advanced stress analysis. 4.5 Advantages In addition to human expert’s time being saved and used for finer decisions, the advantage of KBS application is to be able to code specialized know-how such that the knowledge base turns out to be a significant expertise from senior experts. There are significant advantages in its application in terms of saving engineering man-hours, increasing productivity, and being useful in training new technical experts. Furthermore, the medium extent of KBS deployment should deal with high-level decision-making of keen expert’s know-how. This KBS for pipe stress analysis together with similar tools for strategic areas is a key aspect in augmenting productivity, preserving, and incrementing organization technical knowledge and being strategic in promoting sustainability of engineering services organizations. Decision-making on pipe stress analysis enabled by knowledge-based systems 273 5 Ongoing technologies for decision-making 5.1 Ontologies and business process languages Nowadays, a competent method to represent information useful to model, design and then implement computational systems is the ontologies. Ontologies provide the elements to precisely define the entities and relationships among entities in certain domain. Ontologies set the methodology and basic elements to construct a representation language in such a domain. By using ontologies, decision tools of a language devoted to make clear communication among a community of users can be constructed [32]. The incorporation of ontology engineering tasks is a must in knowledge-empowered organizations [25], as is the case in engineering companies related to the energy sector. According to ontologies classification, the knowledge ontologies describe domains and facts that characterize the system, whereas the service ontologies depict the abilities each computing module (agent) provides. There are interaction ontologies specifying the protocol attending declarative or actionable message interchange as well as the shared understanding of information contents. The interoperability ontologies specify the layers making communications and coordination (collaboration or cooperation) among heterogeneous applications. The accurate description of structures and pieces alongside a process entails the usage of ontologies. The ontologies-based representation allows a smart design of complex information system, as illustrated in [18]. This ontological description sets easier process integration from structural, functional, or teleological perspective. For our purpose, the PSA ontology embraces description of the plant of processes entities like temperature, diameter size, and connected equipment. Complementarily, the ontological description of physical and chemical cause–effect relationships among the mentioned entities describe the functional and teleological facts regarding pipe stress analysis. Some KBS rules are the computational programming of these physical/chemical relations. There are teleological relations pertaining to the experts’ know-how that are not being programmed yet, and could be considered in further ontological descriptions. The along dynamic has an actionable modeling through business processes modeling language [10] that extends the workflow language. The workflow language models the automation of the whole and a part of business process: according to procedural rules, tasks information is passed between participants for action. In actionable perspective, the sequence/parallel distribution of subprocesses and tasks are fine modeled such that business process modeling allocates a quick management of the dynamic and teleological relationships during engineering processes, complementary to the ontological description of related entities. Furthermore, BPML places manage technical information about the process, on participant’s profile data, as well as on the workflow process documents, namely input/output information, used/provided for/from the parts of process PSA analysis herein. BMPL family extends deployment facilities, e.g., the business process execution language for Web services (BPEL4WS). BPEL4WS, 2006, specification is a recent option enabling Web services standard for composition such that it allows creating complex processes by wiring together activities including data manipulation, correlation, fault handling, compensation, and begin/end of structured activities alongside processes. 274 M. Alvarado et al. 5.2 Autonomous agents and rationale The autonomous agent systems technology enables social interaction so as to manage input/output data through sequential or parallel steps, without a whole centralized control. Autonomy on computing, meaning the capacity of a system’s module to self-manage the processing and answer (output) attending environment requirements (input), is a distinctive characteristic of agents as components of the new generation of distributed systems. Autonomy entails agent’s processing of incoming messages or effects from external actions, such that own-managing on time and conditions of reply is practiced. The autonomous task executions to agent’s goals-achieving are coordinated on the base of individual or shared agent’s plan actions. In addition to autonomy, the abilities defining an autonomous agent (AA) are pro-activity so that agents plan and act to reach their own no-externaldemanded goals; knowledge-reasoning process: information an agent has about own and the system’s states that are processed through internal inference mechanisms (agents mental structure). The Workflow Automation Agent-Based Reflective Process (WARP) approach proposes a methodology and frame to implement complex process attending the structural, dynamic and functional facets [8]. Multiagent systems supported on Web services cross through the whole activities and actors through the organized workflow. Actually, business process specification gets an effective autonomous-agents-based decision-making deployment [33]. Systems of autonomous agent’s self-managing participation enable next KBS generation based on the use of ontologies that organize and do context making meaningful information [34]. Recording the know-how and the best lessons learned is fundamental in order to be able to reuse previous information and experiences in decision-making. Questions to attend, on one hand, involved what to record from the partial results; on the other hand, how to ensure that the resulting records are not only easily retrievable and in a format that allows its reconstruction by future users, but also amenable to be processed by a computer in order to achieve a degree of automation in an effective fashion [2]. 6 Conclusions Engineering decision-making mediated by KBS for pipe stress analysis is an opportunity to contribute to the effectiveness, the increase in productivity, the enhancement of technical knowledge, but as a major strategy to promote sustainability of engineering services organizations. This area of engineering has been focussing on AI techniques since the late 1980s with variable interest, but we believe the current advances both in the engineering discipline (including knowledge and commercial software) and in AI, methods are in a strategic position as to be combined to enhance the productivity in engineering practice. Acknowledgements This work was done under the collaboration of Project Engineering Direction and PIMAyC (Applied Mathematical and Computing Research Program), both from the IMP. We would like to express our gratitude to the anonymous reviewers for their comments and suggestions. M. Alvarado and M. A. Rodrguez-Toral would like to thank the Mexican National Researchers System for supporting their research activities. Decision-making on pipe stress analysis enabled by knowledge-based systems 275 References 1. Abou-Ali MG, Beltagui SA (1995) Expert system for the integrated design of heat exchangers. Proc Inst Mech Eng, E: J Process Mech Eng 209(E1):27–39 2. Alvarado M, Bañares-Alcántara R, Trujillo F (2005) Improving the organizational memory by recording workflow, decision-making and rationale. J Pet Sci Eng 47(1/2):71–88 3. Ashraf O, Chin M, Hollings J et al (1985) Interactive pipe stress analysis on microcomputers. Am Soc Mech Eng Press Vessels Piping Div (Publ) PVP 98-5:121–124 4. Barton P, Pantelides C (1994) Modeling of combined discrete/continous process. AIChE J 40:966–979 5. Bañares-Alcántara R (1995) Design support systems for process engineering I: Requirements and proposes solutions for a design process representation. Comput Chem Eng 19(3):267–277 6. Bañares-Alcántara R, King J (1997) Design support systems for process engineering. III: Design rationale as a requirement for effective support. Comput Chem Eng 21(3):263-276 7. Bayer B, Marquardt W (2004) Towards integrated information models for data and documents. Comput Chem Eng 28(8):1249–1266 8. Blake B, Gomaa H (2004) Agent-oriented compositional approach to service-based crossorganizational workflow. Decis Support Syst 40(1):31–50 9. Bogarin Jose AG, Ebecken Nelson FF (1996) Integration of knowledge sources for flexible pipe evaluation and design. Expert Syst Appl 10(1):29–36 10. Assaf-Arkin I (2002) Business process modeling language available at: http://xml. coverpages.org/BPML-2002.pdf 11. Sanjiva W, Curbera F (2006) Business process with BPEL4WS: concepts in business processes, available at: www-128.ibm.com/developerworks/webservices/library/ws-bpelcol1/ 12. Brice A, Johns W, Castell C, Banares-Alcantara R, Leboulleoux P, Sellin L (1998) Improving process design by improving the design process. In: AIChE annual meeting, Miami, FL, pp 15-20 13. Chatterjee M, Unemori A, Kakaria A, Jain D (1992) Integrated pipe stress analysis/support pattern selection/support design CAE system. In: Proceedings of the 1992 ASME international computers in engineering conference and exposition, vol 2, pp 233–240, San Francisco, CA R 14. COADE (2001) CAESAR II pipe stress analysis software. Product Catalog 15. Diab YG, Morand D (2003) Risks analysis for prioritizing urban sewer rehabilitation: a decision support system. New Pipeline Technol Secur Saf 1:610–620 16. Durkin J (1996) Expert systems design and development. Prentice Hall, Englewood Cliffs, NJ 17. Ebecken NFF, Geymayr JAB, Gottgtroy MPB (1992) Expert systems development for evaluating the physical integrity of process equipment in the petroleum industry. In: Proceedings of the 17th international conference on applications of artificial intelligence in engineering (AIENG’92), Waterloo, Ontario, Canada, pp 335–344 18. Gusikhin O, Rychtyckyj N, Filev D (2006) Intelligent systems in the automotive industry: applications and trends. Knowl Inf Syst, 12(2) 19. FIPA iterated contract net interaction protocol specification, Foundation for Intelligent Physical Agents, available at: http://www.fipa.org 20. Graham-Jones PJ, Mellor BG (1995) Expert and knowledge-based systems in failure analysis. Eng Fail Anal 2(2):137–149 21. Jellesed RH (1989) Developing an expert system to link the piping designer to computeraided design. Am Soc Mech Eng Press Vessels Piping Div (Publ) PVP 177:49–54 Presented at the quality use of the computer: computational mechanics, artificial intelligence, robotics, and acoustic sending, Honolulu, HI 22. Jennings NR (2000) On agent-based software engineering. Artif Intell 117:277–296 23. Kang S-S, Myung S, Han S-H (1999) Design expert system for auto-routing of ship pipes. J Ship Prod 15(1):1–9 24. Kim Y-J, Suh M-W, Seok C-S et al (1996) Development of expert system for nuclear piping integrity. Am Soc Mech Eng Press Vessels Piping Div (Publ) PVP 323(1):207–215 25. Kotis K, Bouros GA (2006) Human-centered ontology engineering: the HCOME methodology. Knowl Inf Syst 10(1):109–131 276 M. Alvarado et al. 26. Lindley DW, Yow JR, Knott R (1989) Using integrated pipe stress/support analysis to establish extra safety margin. Am Soc Mech Eng Press Vessels Piping Div (Publ) 182:97–102 27. Marra J (1995) Use of knowledge-based engineering in compressor rotor design. In: American Society of Mechanical Engineers (Paper), paper no: 95-GT-384, 11 pp Proceedings of the international gas turbine and aeroengine congress and exposition, Houston, TX 28. McLean A, Young RM, Moran T (1991) Questions, options and criteria: Elements of the design space analysis. Hum Comput Interact 6:201–250 29. Mezher T, Abdul-Malak MA, Ghosn I, Ajam M (2005) Knowledge management in mechanical and industrial engineering consulting: a case study. J Manag Eng 21(3):138–147 30. Mohiuddin AKM, Kant K, Sangal R (1996) ESTOWER: an expert system for the thermal design of wet cooling towers. Eng Appl Artif Intell 9(2):185–194 31. Morrill JP, Wright D (1989) Method for reasoning by analogy in failure analysis. J Vib Acoustics Stress Reliability Des 111(3):306–310 32. Morbach A, Yang AD, Marquardt W (2007) OntoCAPE—a large-scale ontology for chemical process engineering. Eng Appl Artif Intell 20(2):147–161 33. Chen M-Y, Chen A-P (2006) Knowledge management performance evaluation: a decade review from 1995 to 2004. J Inf Sci 32(1):17–38 34. Nemati H, Steiger D, Iyer L, Hershel R (2002) Knowledge warehouse: an architectural integration of knowledge management, decision support, artificial intelligence and data warehousing. Decis Support Syst 33:143–161 35. Ohtaki S (1995) Thermal stress analysis of pipe bends by the finite element method. Am Soc Mech Eng Press Vessels Piping Div (Publ) PVP 305:417–423 36. O’Keefe RM, Preece AD (1996) The development, validation and implementation of knowledge-based systems. Eur J Oper Res 92(3):458–473 37. Peng LC (1978) Stress analysis methods for underground pipelines. Pipe Line Ind 47(5):65– 74 38. Peng LC (1979) Toward more consistent pipe stress analysis. Hydrocarbon Process 58(5):207–211 39. Prassl WF, Peden JM, Wong KW (2005) A process-knowledge management approach for assessment and mitigation of drilling risks. J Pet Sci Eng 49:142–161 40. Qin X, Regli WC (2003) A study in applying case-based reasoning to engineering design: mechanical bearing design. Artif Intell Eng Des Anal Manuf: AIEDAM 17(3):235–252 41. Robleto RA, Tseng MS (1989) Diagnosing high-energy piping problems with an expert system. Am Soc Mech Eng Press Vessels Piping Div (Publ) PVP. 169:137–142 42. Robleto RA (2002) Modeling underground pipe with pipe stress analysis program. Am Soc Mech Eng Pressure Vessels Piping Div (Publ) PVP 440:131–136. 43. Robleto RA (2004) Reduction in stresses shown in piping programs in large diameter pipe branch connections by applying flexibilities computed by shell finite element analysis. Am Soc Mech Eng Press Vessels Piping Div (Publ) PVP 447:55–59 44. Sapuan SM (2001) A knowledge-based system for materials selection in mechanical engineering design. Am Soc Mech Eng Press Vessels Piping Div (Publ) PVP 22(8):687–695 45. Sawa T, Ogata N, Nishida T (2002) Stress analysis and determination of bolt preload in pipe flange connections with gaskets under internal pressure. J Press Vessel Technol 124(4):385– 396 46. Sheremetov L, Batyrshin I, Martinez J, Rodriguez H, Filatov D (2005) Fuzzy expert system for solving lost circulation problem. In: Proceedings of the 5th IEEE international conference on hybrid intelligent systems, Rio de Janeiro, Brazil 47. Subrahmanian E, Konda SL, Levy SN, Reich Y, Westerberg AW, Monarch I (1993) Equations aren’t enough: informal modelling in design, AIEDAM 7(4):257–274 Decision-making on pipe stress analysis enabled by knowledge-based systems 277 Author Biographies Matı́as Alvarado is a Research Scientist at the Centre of Research and Advanced Studies (CINVESTAV-IPN, México). He got a Ph.D. degree in computer science at the Technical University of Catalonia with a major in artificial intelligence. He has a B.Sc. degree in mathematics from the National Autonomous University of Mexico. His interests in research and technological applications include knowledge management and decision-making, autonomous agents and multiagent systems for supply chain disruption management, concurrency control, pattern recognition, and computational logic. He is the author of about 50 scientific papers, the Guest Editor of journal Special Issues on topics of artificial intelligence and knowledge management for the oil industry, and an Academic, invited to the National University of Singapore, Technical University of Catalonia, University of Oxford, University of Utrecht, and Benemérita Universidad Autónoma de Puebla. Miguel A. Rodrı́guez-Toral is a Chemical Engineer educated at the University of Edinburgh, U.K. (Ph.D.), UMIST, U.K. (M.Sc.), and UNAM, México (B.Sc.). He has 13 years of work experience at the Mexican Petroleum Institute (IMP) in the areas of engineering design of heat transfer equipment, cogeneration, and process engineering for the oil, gas, and petroleum refining industry. He is currently the topside leader of the Deepwater program at the IMP. He has interest in the applications of mathematical optimization and knowledgebased systems for the solution of process engineering and energy efficiency design problems. Armando Rosas Elguera is a Civil Engineer working at the IMP. He has 27 years of experience as a Specialist in flexibility and support of critical piping systems for the process industry. In 1979, he was a piping stress and flexibility Specialist, then an Office Head of piping flexibility, Coordinator and Representative of the IMP in the Laguna Verde project (a nuclear power plant in Mexico). He was also the Head of the pipe stress analysis department from 1994 to 1998. Currently, he is a Researcher in the applications of pipe stress analysis. He has deep practical experience in pipe stress analysis for nuclear power projects, for process and power plants involving all the different phases of engineering projects, from engineering design to plants start-up and operation. 278 M. Alvarado et al. Sergio Ayala got a B.Sc. degree in civil engineering from the Mexican National Polytechnic Institute (IPN). He is now retired from the IMP. He has more than 30 years of industrial experience gained at the IMP in the area of pipe stress analysis of process plants. He has extensive practical experience in the engineering design and technical advice during start-up and operations of piping systems for the upstream and downstream sectors of the Mexican petroleum industry. He is a Senior Specialist in pipe stress analysis. He has interest in the applications of computer science for the implementation of a corporate memory in his area of speciality.