cs.SEAug 19, 2026

Building real-time digital twin instances with Function+Data Flow: user evaluation and extension for iterative pipelines

Authors: Eduardo de Conto, Blaise Genest, Arvind Easwaran, Nicholas Ng, Shweta Menon

Organizations: College of Computing and Data Science, Nanyang Technological University, 50 Nanyang Avenue, Singapore, 639798, Singapore. · CNRS, IPAL Lab, France.

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Digital twins (DTs) increasingly leverage artificial intelligence (AI) and machine learning (ML) pipelines, both to build real-time DTs from high-fidelity simulations and to instantiate them with historical data. However, engineering these pipelines remains largely ad-hoc: pipelines are hard to specify, validate, and reuse, with scarce dedicated tooling. Function+Data Flow (FDF) addresses this by defining a visual domain-specific language (DSL) that represents functions (ML models) explicitly, enabling their composition and reuse. We implemented FDF in DesCartes Builder, an integrated modeling environment supporting FDF-based DT synthesis and validation. In this paper, we report on an empirical user study evaluating whether FDF and DesCartes Builder can make AI-based DT development more accessible and reliable. Participants implemented a representative real-time DT prototype within DesCartes Builder, and we measured perceived usability and feature adequacy through quantitative and qualitative measures. Our results indicate that DesCartes Builder and FDF achieve a good level of usability across a broad range of potential users, and particularly for the intended audience of domain experts. The study additionally surfaces concrete strengths and areas for improvement of both the tool and the underlying FDF framework. Informed by these findings, we propose H-FDF, a Hierarchical extension of FDF supporting iterative and modular pipelines, enabling the formal specification of more complex DT pipelines such as dual training. Our findings suggest that integrated, model-driven platforms are a promising direction to transform AI-based DT engineering into a disciplined modeling practice.

Figures & tables

Figure 1: Summary of DT engineering phases. FDF De Conto et al. (2024) and DesCartes Builder de Conto et al. (2025) focus on the second and third phases of fast DT learning ( Φ 2 ) and DT instantiation ( Φ 3 )
Figure 2: Pipeline to design a real-time DT prototype for material strain prediction (extracted from De Conto et al. (2024) )
Figure 3: Minimal FDF pipeline implemented in DesCartes Builder . The number next to the ports indicates a possible inferred implicit type
Figure 4: Example of FDF’s implicit type warning as implemented in DesCartes Builder when attemping to connect y to the function reduce that expects an inputs with same implicit type as X in the FDF pipeline of Fig. 3
Figure 5: Comparison of FDF pipeline in DesCartes Builder (top) and equivalent KNIME pipeline (bottom)
Figure 6: Depiction of the DesCartes Builder architecture
Figure 7: DesCartes Builder user interface showing the canvas ( C 1 ) , the library selector ( C 2 ) , and the parameter editor ( C 3 ) . A simplified implementation of the material strain prediction real-time DTP (Fig. 2 ) is displayed. It uses a split_data block to partition displacement ( displ ) and plastic strain ( eps ) into training ( _train ) and test ( _test ) sets. The Trainer then learns a function predict from ( displ_train , eps_train ), which a Processor then applies to displ_test to estimate eps_p . Last, a score box compares eps_p against the test data eps_test , outputting metrics such as NMRSE and R2R^{2} . The function predict is exported to constitute a real-time DTP.
Figure 8: Implementation of the material strain reduced-order model (ROM). It uses two Coder boxes to learn the dimensionality reduction basis, two Processors to obtain the reduced displacement ( displ_red ) and reduced strain ( eps_red ), and a Trainer to learn a surrogate model from displ_red to eps_red . Evaluation is performed using a score and a sensitivity_analysis Processor box (adapted from de Conto et al. (2025) )
Figure 9: Testing pipeline for the material strain ROM
Figure 10: Validation results for the material strain prediction use case
IDShort label (positive only)
Q1Q_{1}Intended frequency of use
Q2Q_{2}System simplicity
Q3Q_{3}Ease of use
Q4Q_{4}User independence
Q5Q_{5}Functions well integrated
Q6Q_{6}Design consistency
Table 1: Questionnaire for DesCartes Builder usability (based on the SUS questionnaire Brooke (1996) )
Figure 11: Participant demographics in the DesCartes Builder user study
Figure 12: Box plot with the normalized SUS score (0 - 100 scale, higher = better). Black line: median (72.5); dark red cross ( ×\times ): mean score (68.9); dashed line: unnaceptable score (50)
ToolSUS ScoreReference
Simulink76.4Ferrari et al. (2021)
UPPAAL61.7Ferrari et al. (2021)
Excel56.5Kortum and Bangor (2013)
Extremo70.0Mora Segura, Ángel and De Lara, Juan and Wimmer, Manuel (2024)
DesCartes Builder68.9De Conto et al. (2024) ; de Conto et al. (2025)
Table 2: Mean SUS Scores for selected tools
Figure 13: Normalized SUS item scores (0-10 scale, higher = better). Error bars: standard deviation; diamonds: medians; dashed line: unacceptable score (5.0)
Group ( nn )Mean (SD)95% CI
Technical (15)72.8 (12.1)[66.1, 79.5]
Low-technical (10)63.0 (17.9)[50.2, 75.8]
Total (25)68.9 (15.2)[62.6, 75.2]
Note: Welch’s t(14.4)=1.52,p=0.15,d=0.67t(14.4)=1.52,p=0.15,d=0.67 .
Table 3: SUS scores for subgroups
Figure 14: Participant feedback on specific DesCartes Builder features
Figure 15: Visual syntax for the Module box. The Module box encapsulates a subpipeline executed upon its invocation from a higher-level Module
Figure 16: Example of a H-FDF pipeline P1,P2,\hierP_{1},P_{2},\hier
Figure 17: Dual Training pipeline
Figure 18: Comparison of FDF pipeline in DesCartes Builder (top) and equivalent KNIME pipeline (bottom)
Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

IDShort label (positive only)Full SUS item (positive/negative version)
Q1Q_{1}Intended frequency of useI think that I will use DesCartes Builder rather than other methods.
Q2Q_{2}System simplicityI found DesCartes Builder unnecessarily complex.
Q3Q_{3}Ease of useI thought DesCartes Builder was easy to use.
Q4Q_{4}User independenceI think that I would need the support of a technical person to be able to use DesCartes Builder .
Q5Q_{5}Functions well integratedI found the various functions in DesCartes Builder were well integrated.
Q6Q_{6}Design consistencyI thought there was too much inconsistency in DesCartes Builder .
Appendix
Table 4: Complete questionnaire for DesCartes Builder usability (based on the SUS questionnaire Brooke (1996) )
More in this topic

Model-Driven Engineering

See all 8 papers

Explore similar work

CardsList
Sep 15, 2026

High-Fidelity Digital Twin Data Models by Randomized Dynamic Mode Decomposition and Deep Learning with Applications in Fluid Dynamics

The purpose of this paper is the identification of high-fidelity digital twin data models from numerical code outputs by non-intrusive techniques (i.e., not requiring Galerkin projection of the governing equations onto the reduced modes basis). In this paper the author defines the concept of the digital twin data model (DTM) as a model of reduced complexity that has the main feature of mirroring the original process behavior. The significant advantage of a DTM is to reproduce the dynamics with high accuracy and reduced costs in CPU time and hardware for settings difficult to explore because of the complexity of the dynamics over time. This paper introduces a new framework for creating efficient digital twin data models by combining two state-of-the-art tools: randomized dynamic mode decomposition and deep learning artificial intelligence. It is shown that the outputs are consistent with the original source data with the advantage of reduced complexity. The DTMs are investigated in the numerical simulation of three shock wave phenomena with increasing complexity. The author performs a thorough assessment of the performance of the new digital twin data models in terms of numerical accuracy and computational efficiency.

Diana A. Bistrian

University Politehnica Timisoara, Department of Electrical Engineering and Industrial Informatics, Romania

Neural Surrogate ModelingReduced-Order ModelingDigital TwinsComputational Fluid Dynamics
AskPDF
Aug 20, 2026

ExploraTwin, a Non-Profit Research Platform for Digital Twin Simulations

Digital twin simulations show promise, but current empirical evidence suggests that the approach should be tested before being deployed in any particular context. To lower the friction for researchers and practitioners to test and deploy digital twin simulations, this brief commentary introduces ExploraTwin (https://exploratwin.org), an open-access, non-profit research platform for digital twin survey simulations. ExploraTwin supports two modes. In survey mode, researchers can upload a Qualtrics survey file or create a survey within the platform; select an available sample of digital twins; configure and run the simulation, and export analysis-ready data. In panel mode, researchers can assemble a small group of twins for open-ended conversations, document annotation, and moderated, focus-group-style voice discussions. We also developed CroissantTwin, a standardized data format for adding samples of digital twins to the platform. We demonstrate the survey mode workflow by using the platform to replicate 19 experiments on digital twins from the Twin-2K-500 dataset. ExploraTwin's survey execution fidelity is high: 99.6% of 197,000 answer units returned a structurally valid response on the first run.

Naveen Venkat, Yuchen Qiu, Tianyi Peng +2

Columbia University · Columbia Business School

Digital TwinsUser Simulation
AskPDF
Oct 13, 2026

DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models

The development of a digital twin brings together many stakeholders and technical components. We have previously investigated the reporting of the conceptual architecture for these digital twins as an artifact for discussion. However, we find that the activity of modeling the digital twin for visualization and consensus-building can be quite tedious and time-consuming. In this article, we explore the automation of model generation using large language models, targeting the model required for digital twin reporting. Our approach is designed to ingest digital twin documents and output both the conceptual architecture visualization and a reporting page for that digital twin. We first extract from the input the key details aligning with our twenty-one-characteristic Digital Twin Description Framework described in previous work. Then, we generate a description model in the Ontological Modeling Language, representing the digital twin as a constellation of models, data, enablers, and services. We utilize refinement loops with large language models acting as judges and ontological reasoning to increase artifact quality as measured by our metrics. We investigate how architectural and parameter choices affect artifact quality, such as how refinement loops enable local large language models to produce sufficiently high-quality results comparable to larger models. We also detail insights and lessons learned useful for practitioners, such that they can integrate our techniques to accelerate their digital twin reporting or for other model generation tasks. These insights include the use of medium-sized local models (around 20 billion parameters), the avoidance of large language models as judges and clustering for characteristic extraction, and the recommendation of at least two reasoner loops for Ontological Modeling Language refinement.

Kérian Fiter, Angelica Portocarrero, Bentley Oakes

Département de Génie Informatique et Génie Logiciel, Polytechnique Montréal, 2500, chemin de Polytechnique, Montréal, H3T 1J4, Québec, Canada · Facultad de Ingeniería y Ciencias Básicas, Universidad Autónoma de Occidente, Carrera 25 #115-85, Santiago de Cali, 760030, Colombia

Digital Twins
AskPDF