Data consistency, the secret to efficiency gains through automation

Last week at the openBIM forum in Switzerland, I had pleasant conversations on data consistency and its importance. So far, very few people in the AEC industry understood it, but at this trade fair, I was surprised - especially by the younger generation - they crave it, and I approve!?

Why data consistency

Whenever you want to automate a process, data consistency is the prerequisite.?

A lot of the work AEC professionals do aims at data consistency; they call it business as usual. E.g., When a quantity surveyor goes through the plans and colors them by cost item, they create consistent data. Unfortunately, that's time-consuming manual work. So we need data consistency to do our work, and the way to achieve it is to redo the work of others. Consistent data is the first step to automation and, therefore, to efficiency gains in the 10-100 times category!

Alternative ways to achieve data consistency

Manual rework is just one way to achieve data consistency. The next best approach is standards and norming. When we agree that we call every wall an IfcWall and that every IfcWall has an attribute IsExternal with a value True or False, we don't need to discuss if we should call it Exterior Wall, Wall Exterior EX_WALL, or any other combination.?

Usually, we humans are not good at entering data consistently. Fortunately, that's where computers excel - pun intended! So when we can use computers to enrich the data based on some minimal standard, we can achieve much higher data consistency and connected efficiency gains. E.g., We can ask the architect to name rooms, e.g., offices, and classify them by an area category DIN/SIA of any other norm. Very often, this is double work that could and should be automated. Please look at this example to understand why it should be automated and not quality checked.

Checking one attribute with a rule is simple. The rule is that the Name for a room should come from an approved list.?

Now we add a second attribute e.g., Area classification. The rule could be that the area classification code comes from an approved list. But who guarantees that the information between the two attributes is concurrent?

Therefore, checking two attributes is no longer linear development - you don't need two but three quality rules. The first one is to check the room name. The second is to the area classification, and the third is their connection: the more attributes, the more checks.?

Instead of building up these complicated checks,?it's easier to build automated pipelines to enrich models based on minimal consistent data. e.g., We know the room name is an office. With this information, we can automatically add the area classification, surface materials, heating loads, and much more.

With the abstractBIM, we embrace precisely this idea. With minimal input data - the IfcSpaces - we automatically generate consistent models for further use. Therefore, guaranteeing data consistency. Contact me to know more about possible use-cases.

How to get started with automation

The first step is to work on our mindset.?

1. Front loading of processes. Often I encounter the mindset of "activism." We do work, and we must do it urgently because we did not think up front. Unfortunately, most of the time, it's faster to do work quickly and dirty. The moment we have to redo the work, this changes. Now let's be honest, when was the last time you had to do something only once and how often did something change, and you had to repeat your work?
2. Change the Business Model. When we bill by the hour, companies usually have little interest in automating their work. So automation can quickly have an impact on the business model. The secret is from switching to sell value instead of hours.
3. The idea of one software to do it all. CAD/BIM Modelers often want to do everything in one software tool. Unfortunately, most CAD/BIM tools are good at geometric modeling for plan generation. Data management is not their strength. Instead of having one tool, we should start thinking about workflows and how different tools and humans interact.

What's your experience with data consistency, and how to use it for automation?