8 steps to specifying a Digital Twin

Digital Twin Construction

The global market for the Digital Twin market is expected to grow from a predicted UD$3.6billion to around US$50billion by 2026.

Digital Twins can improve efficiency and add value to the operation of our client’s building and infrastructure assets. The idea of a ‘Digital Twin’ is a ‘hot’ issue in the wider industry now, but clarity on the subject is fairly sunk by the heavy marketing of technology firms.

Here we explain how a Consulting Engineer can work with their clients, the Asset Managers, to decide how the ‘Digital Twin’ can be specified to realize real benefit – following an easy and logical eight steps.

Published standards and guides have not yet grasped this fast-developing part of our industry, so here we draw together recent experience to provide this guide. This will get you from pitching, to giving authentic advice, discussing options with your client, and then to the definition of your scope of work for implementation.

How easily we can pitch to deliver a Digital Twin is about our ability to present a method that makes sense to them - the eight stepped approach detailed here does exactly that.

You can download a slide pack at the end of this page.

8 steps to a Digital Twin

Follow these eight steps, that are set out here and explained in detail below.

  1. What is a Digital Twin?

  2. Ten benefits of Digital Twins for Buildings and Infrastructure

  3. Developing the Digital Twin with your client

  4. Cost savings from a Digital Twin

  5. Passive vs Active Digital Twins

  6. BIM for a Digital Twin

  7. What data do we need for the Digital Twin?

  8. Implementation

STEP 1 - What is a Digital Twin?

A Digital Twin for buildings or infrastructure is a digital representation of the physical asset. This will usually be an expanded use of the BIM (Building Information Model) and GIS (Geographic Information System) with added data and methods for information management.

To borrow from our Data Science colleagues, the Digital Twin is to provide a ‘Single Source of Truth’ (SSOT), where each item of data is recorded of edited in only one place.

For Asset Managers and Facilities Managers, it provides an interactive and managed means of recording data about the physical asset in a software model that can be tested dynamically and used to plan operational activity more efficiently. It is the welcome and complete move away from a cumbersome store of ‘as built’ records that are paper based or in multiple digital formats, towards the fully integrated, managed, electronic record.

For Asset Managers and Facilities Managers, it provides an interactive and managed means of recording data about the physical asset in a software model that can be tested dynamically and used to plan operational activity more efficiently. It is the welcome and complete move away from a cumbersome store of ‘as built’ records that are paper based or in multiple digital formats, towards the fully integrated, managed, electronic record.

Definitions for the Digital Twin

There are many, and as a Consulting Engineer what type of buildings or infrastructure that you help your clients with will depend on the right definition for you. This is an area where the standards have not caught up yet. The most important thing is that you and your client are clear on what they need and you can specify and implement it for them together.

Don’t get confused and end up with different ideas, this is a recipe for certain disappointment. At the same time, don’t get stuck with working out beautiful criteria. This is a new thing and you can lead the way with just experience and sensible thinking.

Leading research and advisory firm Gartner define a digital twin as:

"A software design pattern that represents a physical object with the objective of understanding the asset’s state, responding to changes, improving business operations, and adding value."

John Vickers of NASA is recognized for using the term Digital Twin around 2010 and his later work with Dr. Michael Grieves of Florida Institute of Technology provided the first a proper definition:

"The Digital Twin is a set of virtual information constructs that fully describes a potential or actual physical manufactured product from the micro atomic level to the macro geometrical level. At its optimum, any information that could be obtained from inspecting a physical manufactured product can be obtained from its Digital Twin."

This is a very pure definition and to talk about a digital twin of any building or infrastructure as holding “any information that could be obtained from inspecting a physical manufactured product” is probably expecting a little too much for our own application in the construction industry right now.

You can find the full paper "Digital Twin: Mitigating Unpredictable, Undesirable Emergent Behavior in Complex Systems" here.

A more pragmatic definition has been proposed in the UK Cambridge University based Center for Digital Built Britain, proposing that:

A Digital Twin is "a realistic digital representation of assets, processes or systems in the built or natural environment"

This is certainly an easier definition to swallow, for engineers or clients, in that is vague and indicates just the idea that a Digital Twin is a digital model that is, to some degree, a copy of the real thing. Of course, in the complicated world of buildings and infrastructure, common sense tells us that it is the level of detail that is important.

You can find the full "Gemini Principles" paper here.

More than anything, it is the level of detail that will determine success or failure in making the model, whether there is some cost benefit that makes the effort worthwhile. As Consulting Engineers, whether our client will be happy to adopt the twin and it will be what was expected is critical. There is a very messy set of questions and answers to be worked through at the outset to get everyone on the same page, and we work through these step by step, later.

The Digital Twin for those who like maths!

If you are someone that craves a mathematics basis of a Digital Twin, then you may like this approach.

In use, the Digital Twin is mimicking a process where activities at one stage influence activities and outcomes at a later stage. Perhaps it is easier to think of a manufacturing process than buildings or infrastructure to grasp this.

We are interested in the correlation between different events in the process, and it may not always be obvious how one input is influencing an outcome. If we can use the Digital Twin to model the physical and dynamic situation then we may be able to find and address inefficiencies, so we can improve the value of our operations. We may be able to find and examine the risks in the system and work to mitigate them.

Data Scientist Steve Brunton explains this well on YouTube here.

STEP 2 - 10 benefits of Digital Twins for Buildings and Infrastructure

For buildings and infrastructure, the benefits of a Digital Twin increase exponentially with complexity and scale but are real and attractive for most clients. For most clients, the key benefits will be around the following:

  1. The Digital Twin provides a “single source of truth” where information about the systems and the built environment can be accessed efficiently and updated continuously.

  2. Increased certainty of risk management of critical systems.

  3. It enables rehearsal of maintenance activities to reduce downtime, understand performance improvement, increase certainty and safety.

  4. It is easier to plan preventative maintenance and reduce the number of reactive activities, thereby saving costs.

  5. Changes to buildings or infrastructure can be assessed and planned with greater certainty because of improved quality of information.

  6. With care, greater certainty can be converted to efficiency and operational cost savings.

  7. Lower maintenance costs by predicting maintenance issues before breakdowns occur.

  8. Ease in comparing reliability of one element against another, such as the same device may be sourced from different manufacturers and they may not perform equally.

  9. Better tools to plan and manage use of spaces, including COVID related pedestrian flows and flexible use, including the need to manage access for maintenance activities.

  10. Linkage to space management systems, such as ‘hot desk’ and meeting room planners, can ensure that facilities can be fully serviced with technology and disposables as required.

You can see that all these benefits are gained because of some form of improved information (data) management.

3 challenges of Digital Twins

The challenges are about how ready your client’s organization is to realize the benefits of stepping forward and embracing a greatly improved way of doing things. So, be realistic about the following:

  1. Your client’s ambition, and even your ambition as a Consulting Engineer, may be more powerful than the ability to implement the Digital Twin; the client needs to support the change of operational management and not just the new ideas.

  2. Once the Digital Twin becomes the ‘Single Source of Truth’ it also holds all the client’s information, so the system’s data security needs to be as robust as the client’s best standards. This is particularly important if the BIM model data is held in a common data environment that is not held by the client organization. For construction stage the data may be held in someone’s ‘cloud’, but it is a different situation if the building or infrastructure is operational and maybe business critical. Security is a big issue but it depends case to case what the information security issues are. After the Construction stage all the asset information should be in the BIM model. If the BIM model is then handed over to the owner then it can be located on servers that are subjected to the organisations usual information security systems.

  3. For global organizations, there may be quite different standards of digital maturity and digital adoption from one location to another. You will probably start with a new shiny part of the client’s business, but what about when they want to roll out elsewhere?

STEP 3 - Developing the Digital Twin with your client

If we pitch a Digital Twin to our clients without knowing that it will make their life better, we are asking for big trouble.

If you or your team is experienced, maybe even outstanding, at designing new or expanded infrastructure or designing new buildings or refurbishments, you may not be clear on issues of operation. As Consulting Engineers, we do tend to be at some distance from the day-to-day operational management of any assets that we may provide for our clients. We need to recognize this and understand that that client has a role to play and, perhaps, significant insight that can be shared to get the Digital Twin specification right.

Asset Management and a Digital Twin

Asset Management and Facilities Management are stand-alone professions with specialist practitioners who manage assets every day. It is important to recognize the detailed knowledge that your client’s team will hold, even though there may not be so much clarity on strategy.

To get a Digital Twin specification right you need to peak into the Asset Manager’s world and be able to engage with the people for which day-to-day operations is their life blood.

The Institute of Asset Management (The IAM) is a good resource, particularly their guide “Asset Management – an anatomy”, which you can download for free from their site. Find the IAM here.

ISO 19650 has become the UK’s BIM standard and provides some overview of Asset Management principles in Part 3.

The standard project life-cycle is presented in ISO 19650 and also in the Asset Management world in ISO 55000. This representation introduces two new terms for the continued development of design and asset information through the asset life-cycle. These are the Project Information Model (PIM) during design and construction stage. The model is then renamed as the Asset Information Model (AIM) during the operational stage.

The standardized asset life-cycle model, adapted from BS EN ISO 19650‑1:2018 Figure 3

As we have already established, when we are using BIM representations of our projects it is a continuous process of adding, augmenting and revising asset and design data. The PIM and AIM terms just demonstrate the cultural and mindset gap that exists between the design and construction stage and the operational stage.

It is generally accepted that for most building or infrastructure assets, the capital costs (the cost to build things, or CAPEX) is around 20% of whole life costs, with the other 80% spent over the operational life and eventual disposal (the OPEX). Here we are interested in how the Digital Twin will drive efficiency in that OPEX (operational) stage. This is the start of the story for your sales pitch to your client.

Most clients will have in place some form of Strategic Asset Management Plan, which the IAM and ISO 55000 call the ‘SAMP’. ISO 55000, 55001 and 55002 relate to a management system for asset management, which is referred to as an ‘asset management system’. ISO 55001 specifies requirements for an asset management system, while the other standards detail sector-specific, asset-specific, or activity-specific technical requirements or give guide how ISO 55001 should be interpreted and applied.

In the real world, the Asset Manager will have competing priorities and often competing assets, maybe they have many buildings or infrastructure assets. Overall budgets are usually set ‘top down’ by the enterprise, they are under budget pressure and are almost never built from ‘bottom up’.

When different assets are competing the SAMP budgeting is adjusted to reflect real circumstances. For example, new COVID-19 mitigation and procedures will draw funds from other asset management budgets, different assets may have higher priority than others and CAPEX demands may impact OPEX spend.

Any Asset Manager will need to make a strong internal business case to invest in a Digital Twin, so you need to be able to help them to write that story.

STEP 4 - Cost savings from a Digital Twin

If we assume that we can gain efficiency savings by developing and running our Digital Twin, then we have potential to save up to 18% of maintenance cost by moving to more data driven Predictive maintenance. For a large city center commercial office building, this means about a US$ 400k (£300k) a year saved.

If we taking as an example, of a city center commercial office building in London of around 700,000ft2 (or 65,000m2), we can expect a maintenance cost of around US$2.2m (about £1.7m) a year.

You will find more on the cost of building maintenance for various global cities here.

Now we will consider how we can gain efficiency from implementing the Digital Twin.

How the Digital Twin improves maintenance efficiency

Well, the answer comes from the direction that the best maintenance and operations practitioners are heading, and of course, we have to accept that, just like BIM, this begins in the manufacturing sector and eventually moves like a wave across the construction sector.

There are three well recognized strategies for maintenance. Most clients will be familiar with these from the ISO 55001:2014 that covers ‘Asset management’.

As an example, 20 Fenchurch Street, the ‘walkie talkie’ building in London, shown to the left of the image is about 700,000ft2 (65,000m2)

Reactive Maintenance

This was yesterday’s strategy.

The most traditional, where if something fails, we fix it. This is Reactive maintenance and, regardless of what many asset managers will say, this is the primary approach that you will find in buildings and lots of infrastructure.

Preventative Maintenance

This is typically today’s strategy.

An improvement is to try to take actions to spot potential failures and do something to prevent them. This is Preventative maintenance, and a typical example is where lighting lamps or air conditioning air filters are changed on a routine basis, every year etc. The disadvantage in taking this approach is that we replace and discard some items, for example, air conditioning air filters, that are not at the end of their service life - perhaps some way off.

Predictive Maintenance

This is the strategy for today and tomorrow and where your clients need to be heading.

The progressive step is to move to Predictive maintenance, where we can use data to identify and predict where we need to spend our maintenance cash. This gives us the highest reliability, the least disruption (down time) and we use our data to predict what interventions are required.

In a typical case today, in the organizations where asset management is taken seriously, the actual maintenance activities are a mixture of all three. The US Department of Energy suggests that the mix for most applications is around 55% Reactive, 31% Preventive, 12% Predictive (and 2% other) but in the buildings sector only the Predictive element will be almost zero in most cases.

The same study, and others, state that a Preventative maintenance approach potentially costs 18% less than a Reactive approach. Moving from Reactive to Predictive can save a further 12%.

You can easily see that the direction of travel for the industry is that BIM models will be passed from the construction stage of projects to become the base information for a Digital Twin. This will then enable analysis that produces data and results that will inform the detailed maintenance plans.

There is some study of operational costs for typical buildings and infrastructure types, looking at cost per m2 or ft2, the ideas of % Replacement Asset Value (RAV), and the cost of different maintenance approaches here.

You can find the US Department of Energy Operations & Maintenance Best Practices here and there is some more recent work from 2019 at the Proceedings of the International Conference on Industrial Engineering and Operations Management Pilsen, Czech Republic here.

The three models for maintenance: Reactive, Preventative, Predictive

STEP 5 - Passive vs Active Digital Twins

We talk of Digital Twins as a fully dynamic and interactive representation of the physical asset. In fact, in most of the Buildings and Infrastructure, the industry is a long way from achieving this. We must consider, and help our clients understand, the extent to which the Digital Twin can mimic dynamic operation at day one.

At this time, the Digital Twin is more of a data depository, that can help us all to manage assets more efficiently and effectively. We can add dynamic elements to the BIM model that can mimic and simulate some parts of the dynamic system, perhaps where we have a critical part of the system. Effective integration between dynamic software and our more static BIM software is often a challenge now. Your client may be ambitious but perhaps not as ambitious as to become an industry exemplar or leader. In fact, any dynamic functions can run in parallel with the BIM model, they do not have to be part of the same model, but they can pass data between them.

For now, it is better for us to provide a good quality representation of the physical system, as an accurate data depository, the “single source of truth” – this alone is enough to revolutionize the way that our clients will be able to consider and inform our operational and maintenance decisions.

You must share this view with your client, be sure that they understand. There is so much buzz around this subject that it is quite easy to oversell what is realistically possible. It is up to you how you pitch to your client, but please recognize that there is a risk that they will have too high expectations for what can be achieved. Let them know that the Digital Twin is an essential tool to support their asset management and - not yet - an intelligent model.

Normal function at project stages, based on The US National BIM standard Levels of Detail (NBIMS v3)

STEP 6 - BIM for a Digital Twin

The Digital Twin of an asset is a natural step from the extensive adoption of BIM (Building Information Modelling) in the design phase of projects. The BIM for Asset Management part of ISO 19650-3 was published in July 2020. This addresses information management, in the form of a management process, within the context of the operational phase of assets.

If adopted widely, ISO 19650-3 will be the bridge between the design stage BIM models and the data that is managed in the operational stage. This ISO provides a route that helps Asset Managers to get to their requirements for a Digital Twin, but the Consulting Engineer and Asset Managers need to do some significant thinking to work out what they really need. This is expanded later.

Relationship of BIM LOD and the Digital Twin

During design and construction, the detail in our BIM models is described as ‘Level of Development’ (the LOD) by the US National BIM standard (NBIMS v3) and the American Institute of Architects (AIA) and the ‘Level of Information’ in the ISO 19650.

Using the US National BIM standard LOD is enough for the design engineer to convey the requirements, with LOD 100, LOD 200 or LOD 300 for Design Stage and LOD 400 for Construction Stage. This is enough information to build the project, the CAPEX stage. ISO 19650 does align with the RIBA (Royal Institute of British Architects) well established design stages.

When we apply BIM in the operational stage, we are in a world that does not follow a defined pattern and in a story board that lasts over many years. PAS 1192, for runner of ISO 19650, shows this diagrammatically. It draws a distinction between the Project Information Model (PIM) at CAPEX stage and the Asset Information Model (AIM) at operational (OPEX) stage.

We must manage some uncertainty. The CAPEX stage has well defined steps, but the operational stage has lots of undefined surprises! We can handle this by understanding how the client’s organization manages the operational stage.

For OPEX, the level of detail for Operation and Maintenance is usually considered as LOD 500, but this level of detail is not clearly defined in current standards. Until now, LOD 100 to 400 have become well established, with thinking about LOD 500 left for you, as the Consulting Engineer, to clarify.

The diagram below shows the relationship between the LOD levels and the stages of the lifecycle where they apply. There are some important things to notice. The LOD 400 level model is the Construction stage level of detail. But we know that this will only include detail that is required to build the project, and this is not the same as the information for operation. Comparing the LOD 300 design model and LOD 400, the most significant additional detail will be fixings and the structures that will enable the coordinated systems to be combined.

Summary of PAS 1192-3 (ISO 19650) BIM delivery stages including OPEX stage

The relationship of the Digital Twin and BIM LOD at Operational Stage

UK versus US BIM Standards

In the UK, LOD, was ‘Level of Definition’ in the consultation stage PAS1192-2. This is now replaced by the finalized ISO 16950 standard, where the term ‘Level of Information’ is used.

The ‘Level of Information’ is broken down into two parts, being the ‘Level of Model Detail’, defining the detailed the geometry of the BIM elements, and the ‘Level of Model Information’, defining the development and trustworthiness of the structured data. Notice that this draws a separation between the graphical or geometric information and the non-geometric data.

Who will build the BIM LOD 500 model?

We need a model at LOD 500 as the base data for our Digital Twin, but in the usual project processes and standard Scopes of Work, the LOD 500 level model is not well covered. The Contractor’s Scope of Work may include the production of information for ‘As Built’ drawings (also called ‘Record Drawings’) that will form part of the delivery of Operation & Maintenance information at the handover of the project. It is very unlikely that this will be anywhere near the detail to start the building of the Digital Twin.

Depending upon the capability of the contractor and the commercial contract arrangements, it may not be helpful for the contractor to produce this LOD 500 level of detail. It may work better to be produced by another vendor or in close collaboration with the Consulting Engineer and the client’s organization.

To get to the Digital Twin detail there is a potentially complex process for the Consulting Engineer to lead that will establish what is required in the LOD 500 model, and this can only be done by working closely with the client’s operations team. There could be significant time and cost for this process and the scope is not (yet) a standard deliverable.

The new task is to lead and establish what is required in the LOD 500 model, and this can only be done by working closely with the client’s operations team using a framework to decide what is needed. This process is likely to be done together with the facility management team, who may not be identified at the earlier design stage. There are many questions about the detail of asset data that is needed.

Limitations of applying BIM LOD 500

The US BIM Forum has stated that: “The LOD Specification does not address LOD 500 since that LOD relates to field verification and is not an indication of progression to a higher level of geometry or information.” Unfortunately, the lack of definition means that if LOD 500 is specified, perhaps because the client or user wants data for operational management, or perhaps to move towards a Digital Twin, there is a high risk of wasting resource (and money) by modeling too many elements that are not important. This is addressed in detail later.

STEP 7 - What data do we need for the Digital Twin?

Research by the National Research Council of Canada determined that about 87% of running costs for the buildings they studied result from about 16% of the individual asset items. We can safely say that between 10% and 20% of the asset items contribute to 80% to 90% of the operational costs. There is also some good work by The Robert Gordon University, Aberdeen, UK that came to similar conclusions.

So, what about maintenance costs?

It is usual for about 30% of the asset items to use up about 70% of the maintenance costs. I think we already knew that from our own experience, surely?

The Robert Gordon University study can be found here.

How to decide what asset data we need for a Digital Twin

In the CAPEX stage of projects, the geometric information is the most significant in volume. Most of the non-geometric data is for contained in accompanying specifications as part of the design and construction package.

For the Digital Twin we can start with the BIM LOD 400 model from Construction Stage and then add more data to create the LOD 500 version as the basis of our Digital Twin.

I is worth noting that a typical LOD 500 model is only a site verified LOD 400 model, in other words the information contained within is the 'As Built' model, which is usually the LOD 400 model.

Confused? Don't feel bad about it if you are. The question is what should be in the LOD 500 model. What to add?

Defining LOD 500

We need a mechanism or framework that we can use to work with the client team that will allow us to pick and choose what we need, or perhaps it is easier to say that we will discard what we don’t need.

COBie is the acronym of Construction Operations Building Information Exchange. It is a non-proprietary data format for the publication of a subset of building information models (BIM) focused on delivering asset data as distinct from geometric information. COBie was designed by Dr. Bill East while at the Construction Engineering Research Laboratory in 2007, developing from work done in a military environment.

The COBie data format is an established part of the BIM vocabulary and is embraced by the US National BIM standard (NBIMS v3). In the UK ISO 19650-2 states that non-geometric information exchanges in open data formats should be structured to COBie format.

COBie has some strengths in that it can be spreadsheet based, it in integrated as an add-on to widely used BIM packages such as REVIT, and so can reduce the time and effort needed for entering building information manually.

Choosing COBie criteria

COBie provides a defined set of criteria names. Perhaps you can think of these as variables, for each asset type.

COBie in spreadsheet format for an Air Handling Unit

The COBie criteria, in a spreadsheet format, provides a user-friendly format to select which of the many criteria are useful and those which are not so useful. You can access a set of COBie standard spreadsheets here.

Some useful research presented at the 2019 European Conference on Computing in Construction in Greece looked at the application of COBie criteria on some sample buildings.

They studied the criteria available and used a review group to decide whether they were necessary or not. Of course, this cannot be an absolute selection, but it provides some guidance. The work found that about 60% of the criteria were not normally necessary, which tells us that moderation is important when applying COBie to determine the requirements for the Digital Twin.

This study identified that for most of the Building Services criteria set is useful, as you may expect for a dynamic system. The chart shown is the result of the study that you can take as a reasonable guide to how the balance of criteria may be shared across the LOD 500 definition.

COBie chart from ‘BIM LOD for Facility Management Tasks’ by Seyed Hamidreza Alavi and Nuria Forcada Matheu

The steps for deciding the COBie criteria to use is most likely to work best if you:

  1. Review and understand your client’s asset management systems, approaches and capability.

  2. You review the COBie criteria and decide which data items are likely to be of use to your client.

  3. At this stage you can cross check the balance of the criteria used against the findings from the study, as shown in the chart above. This is an easy way to sanity check what you have done and highlight any areas where you have maybe been too detailed or too light; it is just a cross check but a useful check.

  4. Provide these items in some form of written document or schedule and ask them to review. This may work best if you do a simple presentation taking them through the key items and you may need to engage at different levels with different people in their team to cover all the asset items.

  5. The document can then become a schedule for your Digital Twin specification of LOD 500 for the operations stage BIM model.

You can find the full paper ‘BIM LOD for Facility Management Tasks’ by Seyed Hamidreza Alavi and Nuria Forcada Matheu here.

STEP 8 - Implementation and Conclusion

As Consulting Engineers, we need to take care when we talk with our clients about Digital Twins. It’s an exciting area or work for us but it is also an area that has now real delivery model, standards or established methods. Like most new developments, their benefit is certain, but our clients may not be clear on what they need or want or how to deliver it – this is the opportunity for the engineer to lead the way.

The steps set out here provide a strong framework that can be followed easily to bring some process to the specification of a Digital Twin and will allow engineers to quickly review the key issues and risks and then discuss with clients clearly.

Delivering a useful Digital Twin allows us to get closer to the client’s organization though their operational phase, providing the welcome opportunity to extend our services and become a critical friend. Delivering the Digital Twin needs input from the Contractor and their supply chain, the client, and the engineer. Someone must lead and bring together the requirements across these team, with no established model in place – this is a great opportunity for the Consulting Engineer.

How easily we can pitch to deliver a Digital Twin is about our ability to follow a process that makes sense - the eight stepped approach detailed here does exactly that.

You can download slides here

You can click the link to access some of the graphics on Digital Twins in a PowerPoint slide set.

Authored by Paul Lengthorn

Chartered Engineer, MBA, BEng, member of the Institute of Asset Management (IAM) and independent practicing Consulting Engineer

See also ...