Breaking the Proprietary CAD/BIM Monopoly: Why DEXPI 2.0 is the Open Foundation for Industrial Cyber-Physical Twins

J. McKenney

This paper is part of the WG-05-CAD DEXPI body of work, arguing the position case for DEXPI 2.0 as an open alternative to proprietary CAD/BIM formats. It sits alongside WG-05-CAD-DEXPI-Introduction, which gives the technical introduction to the specification, and WG-05-CAD-DEXPI-XML-Metamodel, which documents its XML metamodel in detail; where those papers describe what DEXPI 2.0 is, this paper argues why an open, vendor-neutral standard is the necessary foundation for industrial cyber-physical twins.

Licence: CC BY 4.0. 17 September 2026.

Executive Abstract#

Most industrial facilities, from chemical plants to data centers, keep their engineering drawings locked inside proprietary software from a handful of vendors. That arrangement is a growing liability. Safety analysts, insurers, and cybersecurity teams cannot ask basic questions about a facility's physical risk, because the underlying data cannot be read or queried outside the vendor's own tools.

DEXPI 2.0 is an open, vendor-neutral standard that describes a plant's piping and instrumentation as a machine-readable graph rather than a closed drawing file. It models equipment at three levels of detail, from a generic functional requirement through to the specific asset actually installed.

The economic and risk case follows. Facility owners who keep engineering data in an open format avoid recurring licensing costs and migration penalties, and they give insurers and reinsurers what they need to price risk on evidence rather than uncertainty.

Abstract#

Critical infrastructure, process manufacturing, and high-density computing facilities are trapped within proprietary Computer-Aided Engineering (CAE) and Building Information Modeling (BIM) ecosystems. Closed binary formats and vendor-locked schemas enforced by Autodesk AutoCAD Plant 3D, Revit, AVEVA, and Bentley restrict access to piping, mechanical, and instrumentation models, blocking automated multi-physics simulation, multi-party systems assurance, and real-time cybersecurity graph compilation. This paper presents DEXPI 2.0 (Data Exchange in the Process Industry), an open, vendor-neutral specification grounded in the ISO 15926 series and modern XML/UML architectures, with equipment and property classes drawn from the ISO 15926-4 reference data library, as the sovereign foundation for industrial digital twins. We analyze the technical mechanisms of vendor lock-in, define the mathematical representation of DEXPI topological networks, and demonstrate its three-tier equipment catalog across functional requirements, vendor specifications, and customer-configured assets. By liberating engineering data into computable graph structures, DEXPI 2.0 establishes the physical substrate for verifiable, open-standard cyber-physical twins.

1. The Crisis of Proprietary Walled Gardens in Industrial Engineering#

Modern industrial facilities (such as chemical refineries, nuclear power stations, municipal water networks, and 100 MW hyperscale artificial intelligence data centers) represent the pinnacle of physical systems complexity. The design, procurement, and operation of these assets span multi-decade life cycles. Despite this longevity, the digital blueprints of these facilities remain trapped within proprietary software formats controlled by an oligopoly of software vendors.

In conventional engineering workflows, Process Flow Diagrams (PFD) and Piping and Instrumentation Diagrams (P&ID) are drafted in proprietary platforms such as Autodesk AutoCAD Plant 3D, Autodesk Revit, AVEVA Everything3D, and Bentley OpenPlant. These systems store hydraulic connectivity, valve flow coefficients, pipe schedules, and instrument loop logic inside proprietary relational databases and closed binary representations.

The operational consequences of this lock-in are severe:

1. Information Encapsulation: Engineering metadata is accessible only via vendor-specific application programming interfaces (APIs) requiring expensive annual seat licenses and proprietary runtime environments.
2. Loss of Topological Computability: When engineering models are exported to neutral formats like PDF or standard DWG, the underlying topological graph (how pump PMP-101A connects to check valve CKV-102 and secondary manifold MNF-01) is flattened into visual vectors. The semantic engineering intent is destroyed.
3. Impossibility of Cross-Disciplinary Co-Simulation: Cybersecurity analysts cannot query whether a programmable logic controller (PLC) vulnerability impacts a flammable fluid line. Actuaries cannot compute the physical blast radius of a valve malfunction. Mechanical engineers cannot share live models with digital twin engines without lossy manual transcription.

❝ Plant design in AutoCAD Plant 3D and Revit has crippled cross-discipline collaboration for twenty years. When we design a 140 kW liquid-cooled AI cluster, our P&IDs contain vital hydraulic information: pipe schedules, glycol-water ratios (PG25), valve Cv ratings, pump head curves, and fail-safe orientations (fail-open vs fail-closed). In Revit, that data is trapped in proprietary geometry blobs. DEXPI 2.0 (ISO 15926 / Proteus XML) breaks this lock by serializing the plant as a machine-readable directed graph. If we can map DEXPI's 3-tier catalog (Requirements → Manufacturer Cut-Sheet → As-Built Asset) to digital twins, plant engineers can simulate failure modes without expensive CAD licenses. ❞ Mechanical Engineer / Piping Specialist (Industrial Process Systems Review)

2. DEXPI 2.0#

Architecture, Standards, and Semantic Foundations

The DEXPI initiative was founded by an international consortium of plant owner-operators (including BASF, Bayer, Evonik, and Covestro) in partnership with major CAE software vendors and academic institutions. Its explicit mission is the complete eradication of data exchange barriers across the plant lifecycle.

2.1 Alignment with the ISO 15926 Series and Information Modeling#

DEXPI 2.0 moves beyond early file-translation experiments by establishing a formal Unified Modeling Language (UML) information model mapped onto the ISO 15926 series (Industrial automation systems and integration; Integration of life-cycle data for process plants including oil and gas production facilities).

The series is not a single document, and which part is meant changes what the claim says. ISO 15926-1 states the scope and fundamental principles. ISO 15926-2 holds the data model. ISO 15926-4 holds the reference data library, and that is the part DEXPI actually consumes.

Under ISO 15926-4, every physical entity, functional requirement, and topological connection is classified within a formal Reference Data Library (RDL). DEXPI uses the POSC Caesar Association sandbox and a specialized DEXPI Sandbox, which carries the additional part 4 class definitions DEXPI needs, to define exact semantic classes:

• PlantStructure: Spatial breakdown, process units, and plant boundary definitions.
• Equipment: Apparatus, rotational machines, thermal exchangers, and containment vessels.
• PipingNetworkSystem: Complete fluid conveyance networks.
• PipingNetworkSegment: Discrete pipe runs bounded by equipment nozzles or branching fittings, governed by ISO 10628-2 taxonomy.
• Instrumentation: Measurement sensors, transmitters, digital controllers, and final control elements (actuators).

2.2 Mathematical Representation of Plant Topology#

DEXPI 2.0 formalizes the plant as a directed multigraph $G_P = (V_P, E_P, \Phi_P)$, where:

The edge set $E_P$ represents physical piping connections and electrical/pneumatic instrumentation signals:

Every piping segment $e_k \in E_{\text{fluid}}$ carries an intrinsic thermodynamic property vector:

Because this representation is machine-readable and vendor-neutral, any external simulation engine can compute pressure drops using the Darcy-Weisbach formulation directly from the XML document without launching proprietary CAD software:

3. The Three-Tier Equipment Catalog Concept#

A primary innovation of DEXPI 2.0 is the operational separation between engineering intent, commercial hardware catalogs, and the physical asset deployed on the plant floor. Proprietary tools conflate these three domains into a single proprietary symbol, making automated component substitution impossible.

DEXPI solves this through a three-tier catalog architecture:

1. Tier 1 (Requirements Tier): Captures process requirements independently of commercial manufacturers. Process chemists and thermodynamic modelers specify operating envelopes without knowing which vendor will win the bidding process.
2. Tier 2 (Vendor Equipment Tier): Contains manufacturer catalogs the vendor has signed. Equipment vendors publish their technical cut-sheets directly in standardized DEXPI XML, exposing calibrated pump head curves, motor electrical loads, and valve flow characteristics. The signature establishes who published each curve and when, which is what the owner needs to hold a vendor to a stated duty point years later. It does not establish that the curve is right; that is what the factory acceptance test is for.
3. Tier 3 (Customer Configured Tier): Represents the physical instance anchored in the operational facility. It binds the Tier 1 requirement and Tier 2 vendor model to a physical tag, serial number, field-bus address, and maintenance ledger.

4. Operational Safety, Interlocks, and the Write-Access Boundary#

When industrial data is freed from proprietary silos, plant operators face a critical question: how do we prevent automated optimization tools, machine learning agents, or remote digital twins from compromising physical safety?

In modern process automation, the Purdue Enterprise Reference Architecture (PERA) and IEC 62443 define strict operational zones. Digital twin models operate at Level 3 (Manufacturing Operations Management) or Level 4 (Enterprise Systems). The physical process, however, is controlled at Level 1 (Basic Process Control Systems) and protected at Level 0/1 by Safety Instrumented Systems (SIS).

❝ Operators don't have time to parse software dependency trees during a thermal excursion. The unified model must respect the Purdue Model and IEC 62443 zone boundaries. Mechanical engineers must be able to view their familiar P&ID schematics, while security personnel view vulnerability blast radiuses. Above all, the digital twin must enforce the hard write-access trust boundary: AI and optimization models may observe and simulate, but analog safety instrumented systems (IEC 61511) must hold final physical authority. ❞ Plant Operations Lead (Critical Facilities Operational Reliability)

DEXPI 2.0 explicitly models safety instrumented functions (SIF) and mechanical interlocks as first-class objects:

• Mechanical pressure relief valves (PRV-101) are defined with independent discharge piping segments that operate through mechanical force, with zero software dependencies.
• Fail-safe valve positions are specified as deterministic properties (FailOpen, FailClosed, HoldInPlace).
• Hardwired emergency shut-down (ESD) loops are isolated from supervisory Modbus/BACnet network graphs.

This ensures that while the DEXPI model is fully computable, the digital twin engine recognizes hard boundaries where software optimization is subordinate to physical law and analog safety circuits.

5. Economic Moats, Vendor Independence, and Capital Allocation#

From the perspective of executive leadership and corporate capital allocation, the persistence of proprietary CAD formats is not merely an engineering inconvenience; it is an unhedged corporate liability.

When facility data is stored in proprietary schemas, the asset owner is subject to compound financial friction:

1. Perpetual Licensing Tolls: Software vendors extract economic rent through subscription models, mandatory cloud migrations, and proprietary viewing tools.
2. Revamp and Migration Penalties: Refurbishing a facility designed fifteen years prior often requires spending millions of dollars manually re-drafting paper or PDF drawings because the original CAD software version is obsolete.
3. Insurance and Actuarial Opaqueness: Insurers and reinsurers cannot independently verify plant resilience, forcing the facility owner to pay risk premiums driven by market uncertainty.

❝ Subjective cybersecurity questionnaires are obsolete. When insuring a $1.2B AI datacenter, underwriters want quantitative proof of risk accumulation. By joining BIM and BOM, the digital twin can run Monte Carlo simulations to compute a modeled Single Loss Expectancy (SLE) and Annualized Loss Expectancy (ALE) from stated frequency and severity assumptions, with each assumption named on the face of the result. This allows CFOs to scientifically justify security capital investments (ROSI) and set actuarially sound captive insurance retention layers. ❞ Chief Financial / Actuarial Risk Officer (Capital Allocation & Critical Infrastructure Reinsurance)

When an organization mandates DEXPI 2.0 across all engineering, procurement, and construction (EPC) contracts:

• The facility owner retains absolute, perpetual ownership of the engineering topology in open XML.
• Any qualified software vendor or open-source tool can ingest, simulate, and verify the facility.
• Digital twin models become sovereign corporate assets rather than vendor-locked subscriptions.

6. Conclusion and Future Trajectory#

Proprietary CAD and BIM tools have served their historical purpose as drafting aids, but their closed architectures now throttle the evolution of industrial cyber-physical intelligence. Critical infrastructure cannot remain dependent on proprietary formats that obscure physical risks and prevent algorithmic defense.

DEXPI 2.0 provides the missing foundation: an open information model, aligned to the ISO 15926 series and classed against the ISO 15926-4 reference data library, that treats the process plant as a computable directed graph. By adopting DEXPI 2.0, industrial operators establish vendor independence, unlock automated safety verification, and prepare their physical assets for direct integration with cyber bill of materials standards. The road to the sovereign cyber digital twin begins with the liberation of the P&ID.

7. References#

The technical claims above rest on the DEXPI (Data Exchange in the Process Industry) specification, on the ISO 15926 series and the ISO 15926-4 reference data library it is grounded in, and on the proprietary CAD/BIM platforms contrasted with them: Autodesk AutoCAD Plant 3D, Autodesk Revit, AVEVA Everything3D, and Bentley OpenPlant. The financial and actuarial figures attributed to named roles in the quoted passages are illustrative scenario framing supplied by this working group, not figures drawn from a published source.