May 2024PDECC—Sakthivel (TATA Consulting Engineers)

HP Tagline--PDECCArtificial intelligence for engineering a better tomorrow—Part 2 S. Sakthivel, TATA Consulting Engineers Ltd., Mumbai, India Part 1 of this article—published in the April issue —explored the importance and application of artificial intelligence (AI) techniques in the domain of advanced process design and engineering. This advancement has enormous potential in the creation of engineering drawings. The integration of AI assistants enables engineers to streamline workflows, mitigate manual errors and expedite project timelines.Part 2 focuses on the use of AI for autocorrecting engineering drawings.Error detection and correction. During the preparation of piping and instrumentation diagrams (P&IDs), errors may arise, such as missing or incorrect instrumentation details, improperly sized lines, flawed piping information, deviations from design standards and inadequate review processes. These errors can lead to significant safety hazards, developmental delays, operational inefficiencies and increased costs. The integration of AI in this process can help mitigate these issues, providing engineers with streamlined workflows, fewer manual errors and quicker project completion.Several concepts and methodologies are available to detect and correct errors in process flowsheets or engineering drawings:

Rule-based error detection: In the context of error detection in process flowsheets or engineering drawings, a rule-based approach involves establishing a set of guidelines derived from engineering standards. The detection of errors relies on scrutinizing whether the drawings adhere to these predefined rules. For instance, this could entail ensuring the accurate usage of specific symbols, validating the correctness of line connections and verifying the consistency in labeling throughout the drawings. Designers and engineers can systematically evaluate drawings, promoting adherence to industry standards and minimizing the likelihood of errors in the depiction of complex engineering systems. Rule-based methodologies involve encoding engineering rules for common errors, such as graph patterns, and subsequently identifying and rectifying these errors through graph manipulations.

Pattern recognition: Employing machine-learning (ML) or pattern recognition algorithms helps to identify common errors based on historical data or predefined patterns. ML or pattern recognition can also recognize recurring mistakes in the arrangement of process flow elements or the consistent misplacement of labels. In a recent investigation,¹⁰ a pre-trained YOLO (you only look once) neural network was employed alongside extensive training datasets containing augmented patterns. The network underwent retraining to specifically identify engineering errors in intricate drawings. The resulting model exhibited considerable promise in streamlining manual tasks. The findings indicated the model’s efficacy in analyzing complex, real-world engineering drawings, showcasing its ability to accurately discern patterns within high-level objects. This suggests a valuable application in enhancing the efficiency of error-detection processes within the context of complex engineering documentation.

Semantic analysis: It is important to study the semantic significance of symbols, components and connections within drawings, along with error detection involving an assessment of the logical relationships and constraints between elements. This process ensures the accurate connection of equipment and instruments by validating their appropriate interconnections according to functional relationships.

ML- and AI-based approaches: Operators can utilize ML models, including deep learning techniques, to learn from datasets containing both accurate and erroneous drawings. The model can then predict and correct errors. For instance, a neural network can be trained to adeptly identify and rectify commonplace errors encountered in engineering drawings.

Autocorrection of engineering drawings. In a recent investigation, a novel methodology utilizing SFILES—a text-based notation for chemical process flowsheets—was explored and recommended for the autocorrection of engineering drawings.¹¹ The study revealed that the proposed autocorrection model demonstrated success in rectifying flowsheet topologies. The evaluation was conducted using a synthetic dataset, and the autocorrection model achieved an impressive top-1 accuracy of 80.1%. This outcome underscored the effectiveness and potential applicability of the developed methodology in enhancing the accuracy and reliability of engineering drawings, particularly in the context of flowsheet corrections.¹² The autocorrection methodology for engineering drawings, using transformer models, represents an innovative approach to enhance the accuracy and quality of engineering documentation. Leveraging transformer models commonly employed in natural language processing and computer vision tasks, this methodology adapts their capabilities to the domain of engineering drawings. The transformer model autonomously detects errors in engineering drawings (i.e., flowsheets) and offers suggestions for correction to the user, essentially performing the task of autocorrecting flowsheets. The autocorrection methodology using transformer models can be seamlessly integrated into the engineering workflow. It serves as a valuable tool for engineers and designers, aiding them in maintaining accuracy and consistency in their drawings. The methodology can significantly reduce the manual effort required to review and correct engineering drawings, thereby enhancing efficiency and productivity. FIG. 5 depicts the predictions of the control structure (in blue) of the model promoted with the process flow diagram (PFD) (in black) as input.Sakthivel-Fig-05Embracing AI methods in process engineering offers several advantages. First, it enhances operational efficiency by optimizing processes and resource utilization. Predictive maintenance, enabled by AI, reduces downtime and extends equipment lifespans. Additionally, AI facilitates real-time quality control, ensuring that products meet specified standards. Supply chain optimization, collaborative design and advanced process control are among the many benefits that contribute to improved productivity and cost effectiveness.Mathematical topology description. Mathematical topology provides a powerful framework for describing the spatial relationships and connectivity within complex process systems. By utilizing topological concepts, such as nodes, edges and connectivity, it becomes possible to create a mathematical representation of the underlying structure. This enables a more abstract and comprehensive understanding of the system’s layout, thus aiding in the development of advanced algorithms and models for process optimization and analysis.Data availability for AI methods. The effectiveness of AI methods in process engineering relies on the availability and quality of data. The vast amounts of data generated by sensors, instruments and control systems serve as the foundation for training ML models. With robust data availability, AI systems can learn patterns, predict outcomes and optimize processes. Ensuring data accuracy, reliability and accessibility is crucial to unleash the full potential of AI in process engineering, allowing for informed decision-making and continuous improvement. Takeaways. AI offers significant opportunities to enhance the drafting of PFDs and P&IDs in various ways. Automated symbol recognition stands out as a key capability, where AI algorithms can autonomously identify and position standard symbols used in PFDs and P&IDs. This not only reduces manual labor, but also enhances accuracy, thereby ensuring a standardized representation of components throughout the diagrams. The intelligent routing of process flow is another significant advantage, with AI assisting in optimizing the layout of process flow lines. This intelligent routing ensures efficient connectivity between various components, contributing to the overall coherence and effectiveness of the diagram.Smart annotation and labeling—powered by AI tools—offer a solution to automate the annotation and labeling of components within PFDs and P&IDs. This automation guarantees consistency and accuracy in representing elements throughout the diagram, eliminating potential discrepancies. Additionally, AI contributes to error detection and prevention by performing real-time consistency checks during the drafting process. This capability identifies errors and discrepancies promptly, mitigating the risk of mistakes and ensuring the integrity of the diagram. The incorporation of generative design powered by AI introduces innovative possibilities by suggesting alternative layouts and configurations. Engineers can explore these suggestions that aid in the development of creative and optimized design solutions to enhance the overall design process. These opportunities highlight how AI can significantly streamline and enhance the drafting of PFDs and P&IDs, contributing to increased efficiency, accuracy and collaboration in the field of process engineering. HPACKNOWLEDGMENT The author extends gratitude to Dr. Artur Schweidtmann, Assistant Professor in the Chemical Engineering Department at Delft University of Technology, for his encouragement and guidance throughout the development of this article.LITERATURE CITED

Dzhusupova, R., J. Bosch, H. H. Olsson and R. Banotra, “Pattern recognition method for detecting engineering errors on technical drawings,” 2022 IEEE World AI IoT Congress, June 2022.

Vogel, G., E. Hirtreiter, L. S. Balhorn and A. M. Schweidtmann, “SFILES 2.0: An extended text-based flowsheet representation,” Optimization and Engineering, May 2023.

Balhorn, L. S., M. Caballero and A. M. Schweidtmann, “Toward autocorrection of chemical process flowsheets using large language models,” December 2023, online: https://arxiv.org/abs/2312.02873

Hirtreiter, E., L. S. Balhorn and A. M. Schweidtmann, “Toward automatic generation of control structures for process flow diagrams with large language models,” AIChE, October 2023, online: https://aiche.onlinelibrary.wiley.com/doi/full/10.1002/aic.18259

First Author Rule LineAuthor-pic-SakthivelS. SAKTHIVEL works as the Deputy General Manager in the Technology Team at TATA Consulting Engineers Ltd., Mumbai, India. His expertise focuses on green chemicals, green fuels, energy transition and decarbonization, with a key focus on evaluating emerging technologies for commercialization. Dr. Sakthivel has experience in process engineering, technology analysis, screening and selection, as well as techno-economic assessments, pilot project development and process hazard analysis. His background also extends to basic, applied and market research, as well as powder and science technology. He has authored several articles in both national and internationally peer-reviewed journals.
CoverAd—AxensContentsAd—HoneywellMastheadAd—MerichemEditorial CommentAd—GraceInnovationsAd—Shell MEMMPodcastsAd—AMETEKColumn—Viewpoint (UOP)Ad—Catalyst Trading Company HVColumn—Sustainability (Alazemi, Petrochemical Industries)Ad—IRPCSpecial Focus OpeningSPECIAL FOCUS—Martinez Gacio (Axens)SPECIAL FOCUS—Ditsch (Oleo-X)Ad—GEI Excellence AwardsSPECIAL FOCUS—Keller (Sustainable Transportation Fuels)SPECIAL FOCUS—Vincent (BASF)Ad—CCS Strategy EuropeProcess Optimization—Wan (Trindent Consulting)Ad—WGLCHeat Transfer—Kulkarni (BASF Chemicals India)Heat Transfer—Donnangelo (Emerson)Ad—BoxscorePDECC—Sakthivel (TATA Consulting Engineers)Refinery and Petrochemical Integration—Al Anazi (University of Bahrain)Ad—InstruCalcHydrogen—Nair (Center for Green Technology & Management)Digital Technologies—Gardner (KBC)Ad—H2T CircSales OfficeAd—HP Circ (Back Cover)