Designing an Optimization - Crane Placement

Our model receives two types of inputs:

• The static inputs are defined by the designers, site conditions and characteristics of the building such as layout, shape, special requirements, etc.
• On the dynamic side, we are going to vary the location of the crane described as a point within the interior space of our building (UV Coordinates on a surface) and the boom length will determine the size of our crane.

In order to represent our model into Revit geometry, we must build each floor and identify the Pick-up and Drop-off zones accordingly.

For this example, all floors were modeled as separate types with different materials to visualize each category. For the Drop-off zones, it's important to give a positive height offset to the floor, so it sits on top of the concrete floor and there are no intersections.

Additional elements can be modeled to enhance the appearance of the 3D image that is going to be exported into Generative Design. In this case, we added columns going across all floors to add a realistic touch. However, keep the context geometry light to prevent the studies from taking up too much memory from your computer.

Generative Design works faster when dealing with Dynamo Geometry. To do this, we must import the elements created in Revit to obtain the geometric properties, location, family types and parameters to convert them into solids, surfaces, lines and points that Dynamo can easily modify.

Using the node "All Elements of Category" and then filtering out by Family Name, we can group the types of floors conveniently to then intersect them with the area of influence of the crane. It is important to add "Remember" nodes after converting Revit geometry into Dynamo geometry to ensure that Generative Design keeps these items for all the generations.

After gathering the floors, columns and special areas we are interested in, we can define visualization styles that will be visible in the Generative Design interface. This is particularly helpful when studying individual generations that can be tricky to interpret by just looking at the numbers. Adding transparency and the colors from the Revit model we built a Dynamo abstraction of the problem we are trying to solve:

Once we have the floors, pickup and drop-off zones we can start defining the are where the tower crane will be placed.

For this project, the crane must be located in the interior of the triangular footprint of the building. To obtain this geometry, the index for the ground floor slab must be identified, and once the slab is exploded into surfaces, the horizontal lower one will describe the points for the interior boundary. Using a surface by patch node we can turn the perimeter curves of the inner triangle into a triangular surface that will be our design space for the point.

There's an added complexity when working with triangular surfaces and UV coordinates, as Dynamo simplifies the surface by assuming it is a rectangle. Along those lines, when we vary the U and V parameters randomly from 0 to 1 we might get points that fall outside of the triangular surface. The workaround we suggest is to add a conditional logic that checks for an intersection between the triangular surface and the prospect point. If they intersect, we would return the point resulting from the UV Coordinates, but if there is no intersection we would return a backup point, which for this example is defined as the midpoint of our surface (0.5,0.5).

After solving the intersection issue in the placement point, we can build the cylinder of influence that represents were the crane will be able to operate. To do this:

• We started from the placement point and built a circle on the ground floor using the node"Circle.ByPointAndRadius".
• Then, this circle was translated on the Z direction using the standard height of the crane. Followed by this, we lofted a solid between two profiles that generated the red cylinder representing the crane influence zone.

At this point we have each one of the floor areas: Regular concrete floors (grey geometry), Pickup Zones in the ground floor (green geometry) and the Drop-off Zones on each level (yellow geometry). Using these types of surfaces -- once we exploded the solid floors coming from Revit -- we are ready to intersect geometries and find our areas of interest and get the outputs.

Going back to our problem, we are trying to find the optimal location for a tower crane according to the area that it can service. This area is categorized into regular floor area, pickup area in the ground floor and several drop-off areas scattered around each level according to design drawings and the logistics of the construction process. With this in mind, we want to look at the physical intersection between the crane cylinder and each one of these surfaces. Our optimal solution would be one that maximizes all three. Note that having the crane close to the pickup area might imply that less area is covered on the slab and drop-off category. Here's a compact way to intersect geometries using code blocks.

So far, our model:

• Represents the problem using both Revit and Dynamo Geometry.
• Has geometrical functions applying translations, offsets, transformations and creating solids within our script and we are reporting numeric outputs.
• Offers dynamic inputs — the U and V Coordinates and the Boom Length of the crane.
• Finds the intersecting areas using surfaces from the floors and the cylinder of the crane.

Our outputs contain 3 intersection areas which are:

• the Floor Area (intersection between grey slabs and the red cylinder)
• the Drop-off Area (intersection between yellow Drop-off Areas on each level and the red cylinder)
• the Pick-up Areas (intersection between the green Pick-up area and the red cylinder).

The Generative Design tool generated 20 iterations of our model, and we outlined the highest performing in terms of pickup vs drop-off area.

An interesting behavior shows up when looking at the Parallel Coordinates graph, as there appears to be an inverse relationship between drop-off and pick-up area. Regarding the regular floor area (concrete slabs) we can see from the graph that it behaves very similar to the drop-off area, which could tell us that this area is quite scattered across each level.