entrance dome to PSE Konstancin-Jeziorna - The toughest steel detailing in my life


June 2005. Afternoon. I’m sitting on the sofa in my living room with beautiful sunshine coming in through the window and a glorious blue sky above. Suddenly, the phone rings – an unknown number – so I pick up. A very nice man introduces himself and then asks me if I’m in the business of shop design preparation for steel structures. I am. I am 28 years old and graduated 4 years before. In that time, I worked for the best steel construction company in Wrocław. While there, I gained much valuable experience in Bocad-based shop design. At the time, however, I had been self-employed for a few months, having left my company job to start my own business. That was not MET Construction Engineering (METIB) yet, as that wouldn’t happen until a few years onwards. I was – as they say in the slang of the industry – the ‘workshop guy’. I had Bocad, the knowledge, the experience, the imagination and the tenacity. The man on the phone gets right to the point. I wonder how he got my number. A lot of people in the construction business know each other, so someone must have given him my number and recommended me. His boss told him to find the right person to develop shop drawings for a steel construction project. I asked what kind of structure he had in mind and said I’d be interested to have more details. The structure he was talking about was a dome-shaped building consisting of ribs, each of a different type, and the whole thing looked like a kind of crooked tangerine. The ribs were made of interconnected tubes within a shell structure that resembled the wing of a plane. The ribs are wider at the bottom and tapered towards the top from both sides. The ribs were joined with elliptical spandrel beams. I said ‘It’s an interesting thing and I suppose it’d be doable but I’d need more data.’ The building housed a gallery supported on the ribs connecting the two levels in what looked like an overgrown spiral staircase. I feel a bit overwhelmed but ask the man to send me more drawings anyway so I can make up my mind as to what to do next.

The year was 2005 and Bocad was the only program I knew that worked with spatial models. Some people in Poland would have heard of BIM but I doubt anyone had seen it. Everyone was sticking to 2D AutoCad.

I got the drawings in July. The entire engineering design for the structure had six sheets. A highly condensed data package. I slowly started to put together a plan in my head. Can it be done in Bocad? What kind of question is that? Of course, it can be done if you are clever and know the programme thoroughly! I spent a few days pondering before I finally made a decision. The client accepted my offer. Off we go… just me, myself and I in my little office working away at a slick dome in a faraway town.

Nature of the design

An elliptical dome made of up steel structural members, 17 metres wide, 36 metres long, 14 metres high. Twenty ribs converging along elliptical paths towards the top. The ribs formed the framework of the building. Each rib was different. Some of the ribs were basically slightly modified mirror images of each other. The ribs to one side of the building rested on the foundations and those to the other lay on structure of reinforced concrete five metres above. Each rib comprised two round tubes – top and bottom – bent to the correct shape. The ribs were wider at the base and tapered towards the top. A vertical plate was welded to the lower tube along its length on the inside, and a horizontal bracing plate was welded onto that. From a height of about 11 metres, the tube was cut from the inside and the bracing plate was welded directly into the tube. On top of this, the tubes between them were tied together with diaphragms at 2-metre intervals. The ribs were enclosed on the sides with the shell made of bent sheets 8 and 6 mm thick, respectively. This looked like an aircraft wing. Ventilation openings had been provided in the shell. Inside the ribs and through the base plate ran the ventilation ducts, electrical wiring and sprinkler piping. The widest rib was about 1200mm wide at the base and about 650mm at the top. The ribs were approximately 480 mm thick at the base

and about 250 mm at the top. The ribs joined at the top capped with a welded elliptical box profile sized 300 x 750mm. Externally, the dome was girded with spandrels of circular tubes to form an elliptical shape. The vertical spacing of the spandrels was around 2 metres. The spandrels joined to the ribs with welded studs. The dome was traversed in the middle with a gallery. Coming off from the ribs were metal brackets more than 2 metres long and 22 mm thick. This was overlapped by box profiles forming the gallery structure. The gallery supports were at different levels, as the gallery connected two usable levels from – +3.7m to +4.9m. A complicated affair.

Some views below.

Design approach

I started, as usual, with the axis layout. The next step was to insert all the supporting reinforced concrete pieces. All relevant boundary conditions had to be modelled. This was the only way to ensure that everything works out fine during assembly. I then prepared the batch files in DWG format accordingly so that they could be imported into Bocad. In DWG, the ribs were drawn with the spline function, but after import Bocad broke the spline into rectilinear sections. The only right way to do it. I was working with version 19 from 2005, even though version 20 was available at the time but it had a lot of bugs so I decided to rely on the more stable Unix-based version. Good old Bocad. After importing the rib geometry, I had to produce the feature points and then insert the rib tubes according to these points. Lots of points, lots of ribs, lots of tubes, but it worked. The next step was to cut the bottom tubes. You can pull it off if you do it right. Up to that point, I had relied on feature points to create curves. In the same way, I modelled the bracing plates, ribs and diaphragms. The problems started when I started working on the shell. The shell was to be made of plates of about 500mm each, bent into a curve with a radius of about 900mm, with each sheet different, as the ribs decreased in height, width and thickness. Here, too, I had to be clever. I inserted the plates from one side of the shell to match their decreasing width. Then, in the transverse view, I rotated the shell plates and trimmed them to match the changing rib thickness. Nothing is impossible when you’re clever and know what you’re doing. As I completed that step, I knew the project was perfectly doable. The next step was to work on the vents and the sprinkler system which was to be welded into the ribs. The base plates and bracing ribs went smoothly. The top was a box profile. I inserted the box elements (chords and webs) through the insert element function in Bocad. It was compatible with the DWG version. I also used the same function to handle the spandrels. We need to stop here to common on element discretisation. Bocad breaks each bent profile into a series of inflections. The standard is 12 points per circle. For bars with a diameter of – say – 12mm, this is quite enough, but not sufficient for structures with a diameter of a dozen or more metres. That wouldn’t make the architect happy. I adjusted the number of inflections accordingly so that the pieces came out ‘smooth’ enough and resembled an arch rather than a polygon. The model took me about three weeks to complete. In the meantime, I first released the main tubes for bending. Apparently, at that time there was only one factory in Poland that could bend tubes into such strange shapes. I also released one sample rib so that the design team could verify my methods and make sure the documentation I would be delivering would be suitable for the steel construction factory. The chief architect was also given this documentation to see how the finished product would reflect his vision. We spoke on the phone. He was pleased. Once the model was ready, I started to number the elements and check that everything was right. This is painstaking and responsible work. The software is just a tool, it’s me as a designer who needs to take full responsibility for the finished product.

The programme needs to be checked to make sure it follows your instructions properly. I see today that it’s a common mistake, especially for young people, to trust computer programmes uncritically. This is a big mistake. Never trust a programme completely, always check it to make sure it does what you want it to do. After this step, I proceeded to draw first the individual items and then the prefab items.

There was a multitude of individual components. Especially the rib shell plates. There were somewhat fewer prefab items but they were the most complicated ones I had ever seen. No right angles. How do you draw them so that the people on the shop floor know how to fabricate them? How do you translate complex geometry into flat drawings? My approach is always to imagine that I’m a factory worker and all I have to work with is drawings. Would I manage to cut out and bend all the individual parts correctly? Would I manage to flawlessly assemble a prefab item without any questions? This is another important lesson. Documentation needs to be easy to understand, simple, clear, without understatement or ambiguity. I once heard that if there are at least two ways of doing something, and one of them ends in disaster, that way will always be chosen. That’s true. That is why my design work is always unambiguous and leaves no room for doubt, so there are no problems or disasters. The drawings were, of course, accompanied by a set of lists: materials list, structural list, shipping list, list of drawings, etc. After that came time to produce the assembly documentation. It took me another three weeks to prepare that. Bocad and I did our best and succeeded! Documentation posted. The client was satisfied. The factory started production. The pre-bent tubes for the ribs had been produced and were awaiting assembly into prefab items and reinforcement.

Project progress

I received the input data on 4 July 2005. I released the main tubes for bending on 19 July, i.e. two weeks later. The first sample rib was sent off on 25 July. The other ribs followed on 3 August, and the rest of the shop documentation on 19 August. The assembly documentation was released a week later. This is the pace I’d like to see all my colleagues match.

The tricky bits

As I had my first look at the project, I wasn’t sure if Bocad could handle the geometry. In earlier work I had inserted items according to just a few feature points. Meanwhile, for the rib tubes, dozens of points were needed for each item. The story was the same with elliptical items such as spandrels and the cap. As I mentioned above element discretisation was set to maximum values to make the elements come out as smooth as possible. A very difficult task, which I was not sure about, was the detailing of the rib shell. It contained a great deal of curved sheet metal matching the shape of the rib and additionally bevelled to achieve a reduction in rib thickness. I wasn’t sure if all this could be done in Bocad, but after the first few attempts I knew it would be fine. We just had to come up with the right paths for dealing with the elements one at a time.

Some views from my documentation below.

A tak wyglądała dokumentacja.

.check out our steel structure design service

Production and assembly

The steel construction factory handled the fabrication and assembly. I received some photos of the production (below). Interestingly, I did not receive any question from the production and assembly team. Everything went very smoothly. The executive team did not have any problems. I am pleased, as this is a testimony of perfectly prepared documentation.

Software problems

I had surprisingly few problems with the software given the complexity and number of elements. At one point there was a database issue in Bocad. I accidentally created a model item that was 0 mm thick and could therefore not be displayed or removed. However, with a bit of gumption I pulled it off. That was the only problem with the project and it was self-inflicted at that.

Some number crunching

1,500 individual items, of which 700 items were rib shell plates – it took me three days to draw the shell plates alone. 235 prefab items, 110 tonnes. 6 weeks of work for one person 8 hours a day.


The toughest design job in my life had come to full fruition. No diagrams, no right angles. Never before or since have I done a workshop project as difficult. Later I worked with ellipses, circles and so on but nothing like that.

The building was then encased in glass. Does it look nice? From the outside it doesn’t impress me, but inside it looks amazing.

What does the model look like? You can see it By clicking below. Although it was in 2005 and at the time of writing this article it is already more than 17 years old. All this time I can’t get over how someone conceived such an idea, paid for it and then had it executed.


Link to model (opens in new window).
Some photos of the finished building below. Photos from the urbanity. pl website. Image copyright ©dim84.

I wish for every designer to be able to do the project of their life and climb their Mount Everest. Good luck!

Marek Tomkowicz

Are you interested in this article? Do you have a comment or question? Have you been involved in this project? Enter a comment below.


#steelstructures, #workshop, #steeldetailing, #Bocad, #Schuller&Company, #PSEBielawa, #PSE, #dome, #steel, #structuraldesign, #steelstructuredesign, #METStructuralEngineering, #METIB, #MarekTomkowicz