Let’s Talk About MEP

We often get asked: “Why doesn’t Speckle support MEP?”

And honestly, it’s a fair question—not just for us, but for most AEC platforms out there. Why is MEP so poorly supported in our industry’s digital tools?

The truth? We don’t have a great answer. We’ve focused so far on areas we know well—architecture, structures, infrastructure, computation. MEP has always felt like this complex, vital, yet strangely siloed corner of the industry. We know there are tools people swear by, and we know there’s at least one big gorilla in the room… but beyond that, it’s murky.

So help us out:
• What is the gorilla?
• What’s the niche MEP software you love (or love to hate)?
• What kinds of data matter most in MEP workflows?
• Where does interoperability fall short?
• And what would a great MEP workflow—even a small one—look like inside Speckle?

We’re not making promises. But if it makes sense for you and us, we’re all for re-addressing this imbalance. Now’s your chance to influence where we point our efforts next.

Let us know.

3 Likes

I’ve saved this post to follow what the community will come with this topic

I’m kinda surprised that there is no answer yet..

So, I will try my best to provide as much context and experience I can

For a little introduction, I’m an engineer specialised in complex system (from HVAC to Electricity and services), I’ve work on dozen of projects at different stages (designer, engineer, project management, construction management, maintenance manager…)
Now I’m involved in software architecture around building operating systems

So, HVAC is a vast field… In term of “softwares” you can found “general” ones like Autodesk Insight that is not related to public policies and specialized ones like, in France, Izuba Pleïades or Graitec ArchiWizard that are dedicated to standard calculus and simulations (local regulations like RE2020, standards like the EN 12831)

So, basically, it’s a nightmare in term of interoperability and homogeneity from the engineering part

French approved softwares relies on dozens of databases related to materials, equipments, air quality and so on

Imagine a junction of two walls, this junction is treated as an object : a linear heat sink,
A fixture for an outside heat insulator is a punctual heat sink,
A wall (or any area) is considered as an area heat sink…

That’s for the heat sink evaluation

Then you have appliances and systems,
Again : materials, regulations, treatments…

There are some works around interoperability for those elements : gbXML, BRICKS ontology, BEM (Building Energy Modeling) products

They all have some benefits… but at the end of the day it’s still a nightmare

Graitec even choosed to convert ArchiWizard into a Revit plugin to be more integrated in a BIM workflow (I think that’s because it was easier to work with Autodesk’s SDK than trying to comply with the IFC…)

• What is the gorilla?
I don’t know any. Revit is great for air conditionning systems and to draw piping networks, ultra specialized tools allow to do the sizing and some steps still need some old and dirty spreadsheets..
Some company do their best for HVAC drawings but it still ends some how in a Revit plugin that just draw things a bit faster

• What’s the niche MEP software you love (or love to hate)?
Pleïade is really awesome for whay it can cover (air quality, heat needs, insulation, life cycle assesment and others) but it’s far from being intuitive.
Insight is more like a “one click result” but here we can not certify and use its results.

• What kinds of data matter most in MEP workflows?
Well… it depends :person_shrugging:
MEP is :

  • Heating and cooling area
  • Providing fresh and hot watter to places
  • Renewing the air so the building remains breathable
  • Providing Gas to equipments that needs it
  • Removing waste water
  • Removing smoke from burners
  • Extracting smoke during a fire
  • Puting of a fire with gas like argon, water, foam…
  • Managing special fluids for an hospitals or an industrial process
  • Managing thermodynamics networks with all its specifications

And that’s for the M and the P of MEP…

Now you have Electricity. Quick overview :

  • calculation softwares we use : ElecCalc, Caneco, Etap
  • drawing : schemeElec, SEE Electrical, Eplan
  • lightning : Dialux, Relux
  • construction : mostly Revit but Autocad is still heavily used

Oh I forgot, these are for electrical power distribution systems,

Then you have network (IT) systems (sound, images, voice, data, IoT, IIoT, SCADA…

• Where does interoperability fall short?
Everywhere, we might need to be more specific here

• And what would a great MEP workflow—even a small one—look like inside Speckle?
This is a really goot question but a better one would be : what do you envision Speckle to be in the next 5 years ?

From my point of view, if speckle can reliably manage all kind of engineering informations it will be a dig weight off our shoulders, allowing us to focus on building’s live data and services, leveraging Speckle as the “technical and 3D database” associated as what we could call a digital twin (even if this term is controversary, at least to me)

1 Like

We’re part surprised and part not; MEP has long been the underserved stepchild of digital construction, so maybe it doesn’t even expect to be asked its opinion (OR and I suspect this isn’t the case - perfectly well served by a utopian dream of perfect fit-for-purpose tools). Either way, we’re between a duct and a hard place: “You don’t have anything for MEP,” yet met with crickets when we reach out.

Thank you, Anthea, for laying this all out. To kick things off, here’s what I think we could investigate without baking in rigid schemas, but by leaning on external datasets to copilot designers, based on your contribution:

  • Tool-independent network topology
    Investigate how ducts, pipes, conduits, and other elements are exposed in each MEP tool and map them into a neutral graph format at runtime with no vendor lock-in.
  • External dataset augmentation
    Identify key data sources (regulatory databases, manufacturer catalogues, simulation outputs) and prototype connectors so designers can pull in specs, performance curves, and compliance rules on demand.
  • Copilot-style POC for HVAC
    Build a simple Speckle workflow that ingests your network topology, along with linked external data (e.g., equipment performance from a catalogue API), to provide sizing or validation feedback directly within the designer’s workflow.

Ultimately, our goal is to let your feedback guide us, so even if Civils’ scale networks are first because they are the only discipline regularly reaching out, we won’t lose sight of the full MEP spectrum. Thanks again for keeping us honest!

My experience is that there are quite a few people who know what they want from MEP automation, but have struggled to get any traction as there don’t seem to be many developers for MEP specific tools. Industries like Civil who primarily used CAD and now are adopting Revit and/or other modelling software seem to be on the fast track as a lot of the processes to them are brand new, so they’re able to develop more modern automation instead of being stuck with ‘mature’ MEP template or processes that are harder to change.

The gorilla is that room bounding is so difficult to manage, there’s a significant amount of data that can be stored in Spaces to do MEP calculation or input to MEP design software.

  • This means we spent a massive amount of effort tracing architectural layouts in some PDF editor to get the areas for each room, and then repeated for each architectural version

Another challenge is that Excel exists, a software it’s extremely easy to use, widely accessible, is flexible enough to be used for all sorts of tasks which means people don’t need to a combination of platforms. Where almost all data is stored in calculation sheets, even if the design is done in a 3rd party software it will be collected back in Excel as the ‘database’.

Some of the challenges with MEP although are as applicable to Structural and partially Architecture:

  • Verification - What ‘traceability’ is there so that a Chartered/Professional engineer would feel comfortable tracking the whole process to sign off the data they can’t edit themselves in Excel, even if this is not always done very well in the current manual process, it can be used as a barrier to adopting a new process.
  • Flexibility - Every engineer has their own calc sheet adjusted just slightly from whatever the company/personal standard is for their own convenience. Although I consider this something that could be standardized as long as the automation provides enough convenience to adopt a new method.

The lowest common denominator is one of the biggest challenges, despite all of the discussion around international BIM ‘standards’ they have have little or no impact on the delivery and programme of the design models themselves.

  • The quality of the MEP modelling is fairly proportional to the quality of the architectural and structural model
  • Where the Architect model is lagging (where rooms are not correctly bounded, and ceilings/walls continue to change which affects the MEP reticulation & plantroom setout and delays the final sizing of equipment), or where the Structural design is lagging (which wouldn’t allow you to finalize reticulation routes and/or place penetration openings)

Despite all of these complex models being created, most Heat Load software is quite old, and not amazing for data interoperability.

  • E20 released a new version that took away the ability to import GBXML files making the input process even harder, even though v6 did introduce a model viewer
  • There are newer tools like HVAKR which seems promising but very US centric, or older ones like Camel which just released new Revit integration

A small but I suspect potentially complex workflow would be managing data entry to Revit elements from the browser which I feel like I’ve seen posted about during a Speckle hackathon, Revit schedules are very powerful but having to open the entire model to change one value (even with all worksets closed) is a hard pitch to engineers who do not spend a lot of time in the model.

  • Which could include publishing just data to Speckle and not any geometry?
  • Although there are consideration here about scheduling import/exports for data, how to manage workshared elements that are locked out at the time of import
  • Tracking the change in values and potentially visualizing the change is a task in itself
2 Likes