One label can hide very different purchases. For procurement purposes, it helps to separate three propositions that are easy to confuse: an asset-focused twin, an operational traffic twin, and a visualisation layer.<\/p>\n
A visualisation layer can be a valuable part of a digital twin system. On its own, however, a 3D representation does not necessarily meet the Digital Twin Consortium’s<\/a> current definition: “an integrated data-driven virtual representation of real-world entities and processes, with synchronized interaction at a specified frequency and fidelity”. For contract terminology, ISO\/IEC 30173:2023<\/a> now defines digital twin concepts and types.If the tender does not distinguish the required outcome, bids that solve materially different problems can end up being scored against each other.<\/p>\n Start with what the model represents and what it is meant to do.<\/p>\n An asset-focused twin represents the condition and configuration of physical infrastructure, and may support deterioration modelling, maintenance planning or remaining-life estimates where the underlying data and models allow it. Its inputs can include BIM (ideally exchanged as IFC, whose ISO 16739-1:2024<\/a> edition added roads, bridges and railways), GIS and CAD sources, sensors, inspection records and maintenance history. Ask which asset classes are covered and what supports each one. Then ask how the model handles asset classes with partial or no instrumentation.<\/p>\n For traffic-management decision support, an operational twin can simulate how network conditions respond to an intervention. Ask what it simulates, at what level (microscopic, mesoscopic or macroscopic) and time step, how fast it runs relative to real time on your full network at peak demand, and what it hands back for a given change. If intervention comparison is the use case you are buying, make the supplier demonstrate it. Do not infer that capability from the words “digital twin”.<\/p>\n A visualisation layer does not demonstrate predictive capability on its own. Where a common operating picture is what you need, it can be the right purchase.<\/p>\n A serious digital twin procurement is also a data-integration project. Ask for the input list in writing: sensor types, existing systems, static sources, the update frequency expected from each one, and the interface standard (in Europe, typically DATEX II<\/a>, the CEN 16157 series). Then ask what the twin does when an input goes missing or goes stale. On a live network, inputs will occasionally be missing, delayed or stale.<\/p>\n Do not accept “real-time” as a sufficient specification. The Digital Twin Consortium’s definition turns on synchronised interaction at a specified frequency and fidelity, and that is what to specify: the synchronisation frequency, end-to-end latency, timestamp source, freshness rules, and what happens when an input stops updating.<\/p>\n This question turns capability claims into something you can observe before award. A reference-network demonstration can show generic software capability. It does not establish that the model can reproduce conditions on your network, with your geometry, detector coverage and control plans. FHWA microsimulation guidance<\/a> treats calibration as adjusting model parameters so that the model reproduces performance observed under specific travel conditions, precisely because no single model can be expected to be equally accurate for all of them. So ask for a demonstration built on a section of your own network, with your own data, and write the pass mark into the tender. FHWA’s guidance calibrates one model variant per travel condition against a representative observed day and accepts it only if four criteria hold, including 95% of simulated values within roughly two standard deviations of observed day-to-day variation and two-thirds within one. A well-chosen pilot section may be enough to test the data path, calibration process and proposed outputs.<\/p>\n Three things to watch while it runs. Was the model calibrated against observed conditions on that section, or just configured to look plausible? Can the supplier show you what the model got wrong and explain why? And can someone in the room change an assumption and re-run it while you watch, fast enough to inform a live decision?<\/p>\n Ask what an independent reviewer would actually receive: the simulation engine and its version, the network model, demand inputs, calibration parameters, scenario definitions, and the evidence used to check the outputs against observed data.<\/p>\n The engine is only one of those. The FHWA’s own microsimulation glossary<\/a> defines a model as the combination of the modelling software and the analyst-developed inputs and parameters, and notes that several models can be built with a single software program. Knowing which software sits underneath does not give you the model.<\/p>\n Our own traffic twin uses Eclipse SUMO<\/a>, an open-source microscopic traffic simulation suite originally developed at the German Aerospace Center\u2019s Institute of Transportation Systems and an Eclipse Foundation project since 2018. It is licensed under EPL-2.0 and each release since v1.2.0 has a DOI, so the engine version behind a result can be pinned. An open engine makes one part of the stack directly inspectable. It does not, by itself, make the complete twin reproducible.<\/p>\n A proprietary engine does not automatically prevent independent review either. The procurement question is whether the authority can give a competent third party enough access, documentation and model artefacts to reproduce and challenge a result without depending entirely on the original supplier.<\/p>\n Ask what you will own, what you will have a continuing right to use, and what must be exportable when the contract ends: the network model, the calibration parameters and the historical data, and in what format. Make the supplier prove the export by re-importing it before acceptance. For hosted twins, the EU Data Act<\/a>, applicable since 12 September 2025, adds cloud-switching rules, but does not replace a contractual exit plan.<\/p>\n Much of the project-specific value sits in the network model, the calibration, the live data path and the operational integration. That work is real, and it takes time. Recreating that work after a supplier change can be expensive, which is why the exit terms deserve as much attention as the functional requirements.<\/p>\n A model can drift away from the network it describes. New signal timings, network geometry changes, detector changes and material shifts in demand can invalidate assumptions in the model unless the synchronisation and model-management process captures them. A well-integrated twin can take in some of those changes directly. Others need a deliberate response. Companies in this market fall into four groups: simulation specialists, asset management vendors extending into modelling, visualisation platform providers, and systems integrators who connect a twin to live control systems. Which group suits you depends on whether you need intervention testing, condition monitoring or a common operating picture.<\/p>\n A traffic simulation model can exist as a standalone study model; before synchronisation, the Digital Twin Consortium<\/a> would call it a digital twin prototype. A digital twin adds a defined synchronisation mechanism between the digital representation and the real network, at a specified frequency and fidelity. The underlying simulation engine may be the same; the difference is the operational data connection and its lifecycle.<\/p>\n There is no universal minimum, because the required data follows the use case. For an operational traffic twin, typical inputs include network geometry and topology, control configuration, traffic-state or demand data, and observations for calibration and performance checks. Asset-focused twins require different condition, inspection and engineering records.<\/p>\n Yes, and it should be. A proof of concept on a defined section can show whether the live data path can be built and how operators use the output. We built a live twin in an M1 motorway proof of concept in Hungary for exactly that, calibrated from live field data.<\/p>\n Lillyneir designs and integrates digital twin platforms<\/a> for road authorities and transport operators, from the live data path through to operational integration with the control room. If you are scoping a digital twin procurement, we are glad to talk through<\/a> what to specify.<\/em><\/p>","protected":false},"excerpt":{"rendered":" One label can hide very different purchases, and bids that solve different problems can end up scored against each other. Check what a digital twin contract should settle on synchronisation, model review and exit rights before you award it.<\/p>","protected":false},"author":4,"featured_media":2392,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_seopress_titles_title":"Six questions to settle before you award a digital twin contract","_seopress_titles_desc":"One label can hide very different purchases. What to ask before awarding a digital twin contract, what a demonstration should show, and what to write into it.","_seopress_robots_index":"","_seopress_robots_follow":"","_seopress_robots_imageindex":"","_seopress_robots_snippet":"","_seopress_robots_primary_cat":"","_seopress_robots_breadcrumbs":"","_seopress_robots_freeze_modified_date":"","_seopress_robots_custom_modified_date":"","_seopress_robots_canonical":"","_seopress_social_fb_title":"","_seopress_social_fb_desc":"","_seopress_social_fb_img":"","_seopress_social_fb_img_attachment_id":0,"_seopress_social_fb_img_width":0,"_seopress_social_fb_img_height":0,"_seopress_social_twitter_title":"","_seopress_social_twitter_desc":"","_seopress_social_twitter_img":"","_seopress_social_twitter_img_attachment_id":0,"_seopress_social_twitter_img_width":0,"_seopress_social_twitter_img_height":0,"_seopress_redirections_value":"","_seopress_redirections_enabled":"","_seopress_redirections_enabled_regex":"","_seopress_redirections_logged_status":"","_seopress_redirections_param":"","_seopress_redirections_type":0,"_seopress_analysis_target_kw":"","_seopress_news_disabled":"","_seopress_video_disabled":"","_seopress_video":[],"_seopress_pro_schemas_manual":[],"_seopress_pro_rich_snippets_disable_all":"","_seopress_pro_rich_snippets_disable":[],"_seopress_pro_schemas":[],"footnotes":""},"categories":[8],"tags":[],"class_list":["post-2391","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog-article"],"_links":{"self":[{"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/posts\/2391","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/comments?post=2391"}],"version-history":[{"count":1,"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/posts\/2391\/revisions"}],"predecessor-version":[{"id":2393,"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/posts\/2391\/revisions\/2393"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/media\/2392"}],"wp:attachment":[{"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/media?parent=2391"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/categories?post=2391"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/lillyneir.com\/hu\/wp-json\/wp\/v2\/tags?post=2391"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}What kind of twin is this, exactly?<\/h2>\n
What does it need from your infrastructure?<\/h2>\n
What should a demonstration actually show?<\/h2>\n
Who can independently review the model?<\/h2>\n
What happens when the contract ends?<\/h2>\n
How will you know it is still accurate?<\/h2>\n
\nSo ask what triggers recalibration, what triggers a formal performance check, which performance measures are checked, what tolerances are acceptable (the FHWA criteria above, or the UK DfT’s TAG unit M3.1<\/a> for assignment models, give ready-made templates), and who owns the process under the contract. Then ask what evidence exists that it happened.<\/p>\nFrequently asked questions<\/h2>\n
What companies build digital twins for infrastructure and traffic?<\/h3>\n
How is a digital twin different from a traffic model?<\/h3>\n
What data do you need before starting a digital twin project?<\/h3>\n
Can a digital twin be tested before full deployment?<\/h3>\n