28 December 2011

External timber cladding: the book

Langley Academy, Slough. Foster + Partners. Western red cedar
Timber facades have long been used on low-rise housing in North America and in Scandinavia. Most recently timber cladding is becoming popular in some other countries, Austria and the UK among them. Moreover, timber is nowadays being used as an external finish on medium-rise and non-domestic buildings.

But it is not easy. The main uncertainties related with using timber in facades involve: durability, weathering, dimensional change, corrosion, wind resistance and fire safety. Now, considering this long list, does it mean that timber is unsuitable as an external finish? Far from it. If design intent and construction details are in tune with its characteristics, timber can be a versatile facade material with a unique combination of performance benefits.

Architects in search of guidelines on how to use timber in facades have reasons to congratulate. This post is devoted to a recent book (released in April 2011) whose title says it all: 'External timber cladding: Design, Installation and Performance'. Its authors are Ivor Davies, a researcher from Edinburgh Napier University, and John Wood, professor of engineering at the same Scottish university. The book is more than its authors' baby. It is one of the outputs of a trans-national, EU financed project titled 'External timber cladding in exposed maritime conditions'. The project had inputs from Scotland, Iceland and Norway. More info about the project can be found here.

But this book is much more than the summary of an international study, and it is worth down to the last page if you are interested in timber for facades. In fact, it can be considered as the first true guidelines for timber facade engineering. The authors note very rightly that, during the past decade, facade engineers have tended to ignore timber in favour of more conventional - or more à la mode - materials like concrete, steel and glass. Timber exteriors have been left to architects (general practitioners) and timber specialist suppliers. This has proven risky sometimes, and reductive in most cases. Even more, the main technical standard for facades in the UK (the standard from CWCT) largerly ignores timber, whilst the existing guidance on timber cladding only covers a limited range of topics. This book comes to fill the gap between timber facade construction and facade engineering. It was about time!

The book is structured in six parts, each one dealing with the answer to six fundamental questions, exposed in a sort of 'ignorance pyramid':

What is wood?
How wet does it get?
What effects does it have?
How are these effects controlled?
How do the controls relate to fire safety?
What does all of this mean for facade engineering?

Chapter 1 describes what performance-based design means for timber facades, with a fundamental section on service life. Chapert 2, the top of the pyramid, deals with timber as a facade material, describing its main parameters. Chapter 3 covers moisture conditions in timber facades, and how to predict and to prevent them.

Western red cedar facade with pronounced staining
Chapter 4, as an outcome of the trans-national study, presents the results of site tests conducted on Sitka spruce as a timber cladding. Chapter 5 continues down the pyramid with fungal decay and insect attack. Chapter 6 goes for weathering, and how can we anticipate or respond to weathering in exposed timber cladding. In this chapter we understand why virtually all timber facades in Scandinavia are given an opaque surface coating - good to remember.

Chapter 7 adds to our limited knowledge in dimensional change on wood, and on how / why timber shrinks and moves. Good news for us: movements can be limited and estimated. Chapter 8 goes for corrosion - yes, that of metal fastenings, flashings and brackets embedded in timber. Chapter 9 describes in more depth the structural performance of timber facades - not of structures - which is an often ignored issue. Windloads, robustness of connections, dowel type fasteners and strenght grading are discussed here.

Selection process of timber design
Chapter 10 contains some of the most innovative pages: design for durability. It starts with a decision sequence to aid selection of a timber cladding design from a durability point of view - a must. It then goes down the sequence using the relevant EN standards on timber durability and preservation. It finally relates service life with timber class and use / exposure. Interestingly the authors don't take a side in the discussion pro / against wood preservatives: they present us the arguments in favour and against, so that we can decide case by case - as it should be.

Chapters 11 to 15 deal with fire and timber buildings. The fire triangle, fire testing, fire performance of timber, how to limit external fire spread, the role of air cavities, and a summary of fire regulations in the UK. Finally, a long chapter 16 is devoted to construction details for timber facades. This is an issue largerly discussed in other manuals, but again the authors bring novelty to the case, aided by clear and well drawn details. One of the good points is the treatment given to the junction between heavy (brick) and lightweight (timber) cladding.

The book ends with an updated and interesting list of appendices and references, among them the British and European standards on timber for panelling and external cladding.

Horizontal timber cladding details
In summary: if you are tired of simple and often repeated statements about how timber facades work, and want to know what is really going on and how this should inform your design decisions, this is your book. The authors challenge some of the prevailing assumtions about moisture, its effects and how they are best controlled. New light is shed on how moisture issues affect, and are affected by, the need to ensure that fire safety is fully addressed. And the construction details are based on a combination of new experimental data and a fresh appraisal and synthesis of existing information - they deserve a look and some thought, not just a copy-paste!

Go and buy it. You'll find the link to the publisher at the title of this post. It's not cheap, but it's worth every penny.

17 December 2011

Cupples Products: a Tall Tale of American curtain walling

History is unfair with master builders. Do you know the relationship between these four behemoths of the 20th century architecture: the John Hancock Center in Chicago, the Twin Towers in NYC, the Sears Tower in Chicago and the Hong Kong & Shanghai Bank headquarters in Honk Kong? The relationship is their skin. The curtain walls and external cladding elements of these four buidings were designed and fabricated in the same place: a factory in the plains near Saint Louis, Missouri.

Webb and Knapp Tower in 34th Street, NYC. I.M. Pei
That place had a name that was equivalent to glazed facades during the great decades of the sixties and seventies: Cupples Products Inc. To whom does that name recall anything? Well, ask Norman Foster or I.M. Pei (or Gordon Bunshaft from SOM or even Mies himself if they were alive). Of course, they would answer: Cupples was my curtain wall contractor of choice back then.

That was back then. Today, if you search Cupples in Google you get files from old legal actions and their last address, a suite in Saint Louis. The nearest website is a link to Enclos, the curtain wall company that lastly absorbed Cupples know-how in curtain walling at the turn of the millenium.
There is something to learn from this company, parallel to the growth of US cities in the second half of the 20th century. Cupples was one of the very early entrants into the new industry of curtain walling right after the war ended. Started in 1946 as a manufacturer of residential of window products, the company rapidly progressed into the design development, engineering, fabrication, assembly and field installation of custom curtain wall systems. Cupples became a provider of glazed solutions to architects and builders eager for new facade technology. One of their first client architects was I.M. Pei, for whom they delivered the curtain wall of Webb & Knapp Tower in 34th Street, NYC. This building was built in 1954 for the headquarters of Webb & Knapp , a local tycoon similar to Donald Trump, for whom young Pei worked as an architect during his early career. Pei would remain linked to Cupples as we will see.

U.S. Air Force Academy, Colorado. Cadets Chapel. Skidmore Owings and Merrill, 1959-1963
At the end of the fifties Cupples start working for Skidmore Owings & Merrill, another architect - contractor collaboration that would last long. The U.S. Air Force Academy in Colorado (with its characteristic Cadet Chapel pictured above, finished in 1963) is the most prominent of these early works. It is interesting to note that the chapel has just been refurbished, fifty years after completion. See the SOM webpage for details.
In the fifties and sixties curtain walling was an integrated business. In a way not rivalled by the biggest Chinese suppliers today, Cupples factory included under one roof these production lines: aluminium extrusion, anodizing and liquid coating, cutting, punching, drilling, curving, bending, mechanizing and assembly of curtain wall units. Their designers were able to study and solve the intricacies of a very complex facade; while at the same time they would deliver off-the-shelf simple systems as Horizon, a stick curtain wall very popular up to the nineties. 

Lake Point Towers, Chicago

In 1960 Alcoa, the aluminium giant, buys Cupples and the great story begins for the guys in Saint Louis. The sixties was the era of the sheer towers. New York adopted a zoning resolution encouraging architects to set off their buildings and to enrich land use. The new towers were pulled in from the building line to form landscaped plazas and obtain the maximum permissible sheer height. Soaring from open plazas, aluminium and glass enclosed buildings dominated the US cities. Curtain walling technology also experienced a jump forward: more sophisticated engineering and manufacturing techniques were developed, as pressure equalization and laser technology. Cupples introduced and became a leader in color anodizing.

The sixties were the golden years for Cupples. The list of buildings whose facades were clad by Cupples this decade is simply extraordinary.

The image above, the Lake Point Tower Apartments, is a project of Schipporeit & Heinrich in Chicago finished in 1968. The link to Mies Van der Rohe is obvious, although this is not a Mies's work. Please have a look at this video from 1969 - the third part of three explaining how the tower was built - here dedicated to curtain walling. A real piece of art for us conoisseurs. Cupples worked for Mies in at least two projects: the Pavillion - Colonnade apartments in Newark and the One Charles Center in Baltimore. Two minor projects for the German master but equally interesting.

Another project from the sixties, the John Hancock Tower in Boston, with I.M. Pei & H.N. Cobb (below left). It was actually completed in 1971, but was not opened until 1976. This is the famous (or better infamous) 'plywood building', so nicknamed because of the glass failures it suffered right before completion. This is still the perfect case study for climatic loads acting on insulated glass and how to avoid them - by reducing the stiffness of the bonding between the inner and outer panes of glass. But this is another story.

In San Francisco we find a good example of the decade: the Wells & Fargo Tower, finished in 1966 by John Graham (one of the corporate architects in those days). A very elegant and slender tower, better than most of the SOM buildings in the decade. I am not sure if the vertical shiny cladding is stainless steel or anodized aluminium, but in any case it reflects prosperity and optimism (see picture below right).

Left, John Hancock Building in Boston, I.M. Pei.     Right, Wells & Fargo Bank in San Francisco, John Graham.

Back to Chicago and we find one of the real giants: the 100-storey John Hancock Center, with Bruce Graham and Fazlur Khan (both from SOM) as main architect and main engineer respectively. Completed in 1969, its curtain wall took 2.5 million pounds (1,100 tons) of aluminium and brackets and 300,000 sqf (28,000 m2) of glass.

The John Hancock Center in Chicago by Bruce Graham and Fazlur Khan from SOM, finished in 1969.

During the installation of the John Hancock Center in Chicago, in November 1967, Cupples changed hands for the second time: Alcoa sold it to the metal company H.H. Robertson, and Cupples became Cupples Products Division within Robertson. The new boss was a giant of metal cladding and steel elements for slabs and ceilings, and would soon begin to be well known by their sandwich panels, branded as Formawall. Cupples had a unique position as the only aluminium and glass supplier in the conglomerate, thus allowing Robertson to provide one-stop-shop services for structure and cladding all around the world. Cupples employed 850 persons and was present in curtain walling, aluminium doors, window frames, store fronts, entrances and suspended ceilings at the time of Robertson's take over.
Between the sixties and the seventies, and both in Chicago, these are the two other big buildings clad by Cupples there: the Standard Oil Building and the Sears Tower. The Standard Oil Building, finished in 1972-73, has another interesting story of failure due to the Carrara marble cladding used (not by Cupples) which failed due to bending - a process known as thermal hysteresis. The solution was hard: to remove the whole marble cladding and to replace it with light coloured and thicker granite panels at an incredible cost. 

Left, the Standard Oil Building in Chicago by Perkins & Will with Edward Durrell Stone.  Right, the Sears Tower by SOM.

The Sears Tower, once again by SOM and finished in 1973, will deserve a future post only for itself, so no more comments by now.

We are now well into the seventies. The construction of the World Trade Center in lower Manhattan is going on and Cupples people are busy with fabrication and erection of the two towers, which by the time would be the tallest in the world. Minoru Yamasaki and Associates with Emery Roth & Sons were the architects. Skilling, Helle, Christiansen and Leslie Robertson were the engineers. Tishman was the general contractor. 

World Trade Center towers in lower Manhattan, 1973. Minoru Yamasaki and Leslie Robertson.

Many new design, engineering and construction techniques were required for the structures. The exterior skin was 2-million sqf (185,000 m2) of Cupples aluminium curtain wall. It used a then unique and progressive 'pressure equalizing' design which caused wind loads and pressures to be exterted directly upon the building structure rather than the aluminium skin. The steel frame work forming the exterior wall was installed by hoisting in place 3-module opaque prefabricated units, up to 36 ft high and 10 ft wide (11m high x 1m wide). Horizontal aluminium spandrel units - finished with Alcoa Duranodic bronze - were then spliced onto the adjacent unit. All aluminium profiles and sheets were supplied by Alcoa.

The two buildings included more than 44,000 glass vision units, recessed 10 inches in relation to the external column cladding, achieving some degree of protection from direct sunlight. Another very interesting feature was the anchoring system of the curtain wall to the bottom side of the floor slab. A viscoelastic interlayer between the bracket and the main structure allowed for wind gusts to be transferred but it absorbed part of the dynamic vibration of wind action. It remains to be studied what role - if any - did this viscoelastic shock-absorber have at the time of the planes crash in 2001...

WTC towers during construction. The yellow band between the curtain wall and the steel structure was the area of structure undergoing sprayed fireproofing.

Cupples completed some more interesting projects in the US during the seventies and the eighties. Among them it is worth mentioning two projects of Johnson & Burgee which exemplify the new style of architecture swifting to volumes out of the box or to pure postmodernism. The Garden Grove Community Church in California (aka the Crystal Cathedral) was finished by Cupples in the late seventies. 

Garden Grove community centre (also known as the Crystal Cathedral) by Johnson and Burgee

The Republic Bank Center in Houston, clad with units of pink granite and glass, is a perfect example of the eighties reaction to sheer transparency, which was not bound to last. Another interesting example of big space container made by Cupples is the State of Illinois Center in Chicago, by Murphy and Jahn.

Left: Republic Bank Center in Houston, Johnson & Burgee. Right: State of Illinois Center in Chicago, Murphy and Jahn. 

Republic Bank Center, stone curtain wall horizontal details. Above section @ vision glass, below section @ spandrel

The next - and probably the last - of the great buildings ever to be covered in glass and aluminium by Cupples was not in the US but abroad: the Hong Kong and Shanghai Banking Corporation headquarters in Hong Kong by Norman Foster, finished in 1985. This is a very successful and rare case of project management led by the architect. Foster proposed the Bank to arrange a team of specialists - architects, engineers and contractors - under a sort of design & build contract where architects were the leading partners both during design and construction.
HSBC headquarters in Hong Kong by Foster, 1985: the most expensive corporate building in the world...

The merit that this scheme could go well can be mainly attributed to Foster's bold vision, but the other players - Arup and Cupples among them - were also critical. It is worth mentioning here the role of Phil Bonzon, the engineer from Cupples that led the design and construction of the facade throughout the whole process. Bonzon's sketches were unanimously praised by Foster team members as the only way for them to understand the intricacies of what they were jointly designing. Again, the cladding of this building deserves a future dedicated post if not a whole PhD thesis...

Phil Bonzon's sketches for the HSBC facade design. These details prefigure many industry design features by at least ten years

Nothing can last forever. Cupples lagged behind during the nineties, due not to a specific reason but probably to a number of them. In my opinion the management team did not anticipate that the one-stop-shop model of vertically integrated production was untenable. Cupples people kept for too long a manufacturing structure that was becoming costly and outdated as years passed. The parent company, H.H. Robertson, was also under a similar stress.

Besides, Cupples was not alone in the market. A later entrant to the curtainwall industry had been operating over the decades under the names of Harmon Contract, Harmon Ltd, and finally Enclos Corp, completing many landmark projects. By the end of the 20th century the two companies, Harmon and Cupples, were operating as sister companies under the same umbrella with Harmon focusing on the domestic US market while Cupples tried to keep pace with international operations. Finally, the two companies became one under the same name, that of Enclos. It can be said that Enclos has inherited and continues Cupples history into the twenty first century.

Internal view of the assembly line during the eighties

Why am I interested in Cupples? Because of a personal reason. In 1991 I entered the facade business as project manager with Robertson in Spain, dealing with curtain wall projects and working closely with a number of some great American Cupples colleagues, then busy with projects in our country. I learned the first curtain walling lessons from them. People as Rick Hamlin, now in Trainor Glass; or site managers as James Jutson or Tom Watson, a couple of great chaps whith whom being on site was never boring. Those guys taught me the important lessons. That a facade design is not completed util it is installed - design, fabrication and installation being all part of one same process. That many things can go wrong along the process, and one has to be awake and alert to avoid mistakes and correcting them when things happen. That every step should be checked and re-checked before moving to the next one: it saves time, nobody is perfect. That, in summary, a building and its facade is a lineage of decisions made by a bunch of different people, and success is an outcome of everyone, not of the first one in the line. That being humble on site goes along with being better.

I will never forget those brown helmets with the Cupples name on top. Others will come and build great facades with promising new technologies. But we should never forget those who were so good at opening the trail before us. We owe it to them.

4 December 2011

The Louvre pyramids revisited

Yes: pyramids in plural. This post will compare the main Louvre pyramid (the one we all remember) with the inverted pyramid, less known but equally noticeable. Both are of course part of I.M. Pei's plan for Le Grand Louvre in Paris, commisioned by President Mitterrand as the first of his 'Grand Ouvres'. But as we will see the similarities between the two end right there. The main pyramid epitomises the end of the structural frame era, while the inverted pyramid represents one of the first examples of our time, the supremacy of structural glass.
An unusual view: the main pyramid seen from the top of the inverted one
The main pyramid - together with its three small siblings surrounding it - was finished and opened to public in 1989. The inverted pyramid was not completed until 1993. On the earliest sketches made by I.M. Pei back in 1983 the main pyramid, surrounded by the other three and the pools was already there. The inverted pyramid came slightly later in the plan, around 1985, as a standing point marking the entrance from the underground and the parking.
The main pyramid (centre) and the inverted pyramid (right). The underground entrance takes place from the right.
Pei wanted the main pyramid to be the new entrance to the Louvre, so that people had to wait on the Napoleon court to enter the museum. But this proved not enough to manage the thousands of people arriving every day. Soon predominance was given to the underground entrance, be it from the tube line of from the parking lot. Both coincide precisely at the inverted pyramid, so that it now becomes the first glass feature to be discovered by visitors.

The inverted pyramid between the gardens and the main one

Why are these two structures so different? Aren't both of them just glass pyramids? Well, not exactly. The first difference is their size. The main pyramid has a square base of 35.4m and a height of 21.6m, while the inverted one has a square base of 15.5m and a height of just 7m. Another reason for the two structures to have been designed different is wind load. The main one has to withstand strong wind loads that don't exist in the case of the inverted pyramid. To be precise, the inverted pyramid top side (which by the way is a very low pyramid in itself) only has suction loads due to wind. But the main reason for these two structures to be remarkably different is a fact of evolution: the engineers who designed the main pyramid remained under the old paradigm of glass as an infill; the engineers that were given responsibility to design the inverted one were pioneers of the new paradigm of glass as a structural element.

It has taken me a while to find out who the main actors are in this play in two parts. From the architect's side it is clear : I.M. Pei from Pei & Partners (now Pei Cobb Freed & Partners) was the design principal. Second in charge was Leonard Jacobson, although the design architect in charge of the pyramids was Yann Weymouth. Yann's sketches between 1983 and 1986 are the visual history of the design process, both in general and in its details.

Yann Weimouth and I.M. Pei on site during the main pyramid construction

Studies for tension fastening of the main pyramid. The left version was the selected one. Sketch by Yann Weymouth, April 1985.
Yann, who was fluent in French, lived in Paris between 1984 and 1990. The associate architect (the French local) was Michel Macary.

The 'serres' at La Villette by RFR, 1982-86
Now comes the conundrum of the story: the engineer's selection. The best engineering firm in glass tensed structures worldwide was at the time very active in Paris: RFR, Rice Francis Ritchie. This team of one Irish engineer - Peter Rice - and two British architects (Ian Rithchie was 100% architect; Martin Francis was partly a yacht designer) was created in Paris in 1981. RFR had successfully completed the 'serres' at the Parc de la Villette between 1982 and 1986, so they had to be well known to Michel Macary or to Emile Biasini, Mitterrand's man in charge of the whole project. How then RFR were not given the task of engineering the pyramids? The answer - as much as I can guess - must be found in I.M. Pei's contractual conditions: Pei was made 100% responsible of the design without any interference from French officials or local establishment. This included of course the selection of his engineers - which would be his and not a French firm if he wanted to build his Pyramid as pure as he wanted. It is a real pity that Pei was not right in this point - surely without knowing it at the time.

So the selection for the engineer was made by Pei, and Pei chose the Canadian firm of Nicolet Chartrand Knoll Ltd. What did this firm exactly do at the project? Their role seems to be clear for the large underground concrete structures. Let's not forget that the main pyramid sits atop a 2m-thick concrete slab with large spans. The stair connecting the Napoleon court with the bottom level is also a feat, with its 540º self-supporting curve. As Nicolet Chartrand Knoll refer to the Louvre pyramid in their webpage French version "The scope for the structural engineer, as bluntly expressed by the architect I.M. Pei, was that of building a structure as transparent as technology could reach. Through a close collaboration between the architect, the structural engineer and the other professionals it was possible to reach a successful outcome. Out of about 25 different structures which were studied, one was finally selected ".

Now, letting aside the fact that they studied many options, was the final structure of the main pyramid as transparent as technology could reach by 1986? I humbly disagree, and it seems that Mr Pei was not too impressed by its transparency either. The main point is a conceptual one: this is not the design of a glass pyramid, but the design of a steel pyramid clad with glass. Glass is just filling the space between the stainless steel struts, it is not taking any structural role. As Mick Eekhout - the Dutch structural glass specialist - would put it, the main pyramid is an example of space frame with integrated glazing. The two details below compare a structure of the mid-80s with the main pyramid detail:

Left: a Mero space frame with integrated framing. Right: the top detail of the main Louvre pyramid 

The Louvre solution is fairly more integrated than a space frame with a separate glazing as the Mero structure with glass on top. The outer mullions have disappeared and now a fairly thin aluminium profile is receiving the glass panels. But the glass is simply sealed at four sides with silicone; it is not making part of the structure at all. Suppose we take out all the rhomboidal glass panels: nothing would happen to the structure, the pyramid would remain in place. The pictures from construction period show this quite clearly. Images below have been taken from the book 'I.M. Pei. The Louvre pyramid' by Philip Jodidio, Prestel. It also has great sketches from Yann Weymouth.
Glass being installed at the main pyramid. There was practically no internal scaffolding in this picture. Notice the bespoke cradle to attach the suction cap to the pyramid.

Scheme of the steel structure of the main pyramid. Left: steel tubes in rhomboid shape with cables and stiffeners. Right: the same with glass already in place.
Stages of glass installation on the main pyramid
If the main Louvre pyramid was an American-Canadian design, how come did RFR start working for this pyramid - and yes, much before the inverted one was started? There are two reasons for that, and probably the two were required. The first reason is to be found in the (French) facade contractor: Eiffel Construction Metallique, precisely the same facade contractor that had undertaken the building of the serres at La Villette some years before. The Eiffel guys entered the design team for the Main Pyramid when it was still in the drawing table - remember: 25 models were studied before concluding on one. They must have seen that the Canadian engineers, expert as they were in concrete structures, found themselves a bit lost with the cables and rods of the pyramid. So Eiffel managed to sub-contract RFR as their experts to discuss design subtleties with the Canadians.
Node detail with tubes, bars and cables

The second reason was Peter Rice. He was at the time part of two companies in parallel: Arup - his first employer - and RFR - his new baby. A strong Arup engineering and lighting team, directed by Rice, was working at the same time on the Richelieu wing of the Museum, taking care of the glazing of the three big courts (5,000m2 and 450 tons of steel in total) and of the natural lighting strategies for the whole Richelieu wing. Thus, Rice was already known to and appreciated by the 'proprietaire'. I can imagine Peter Rice inviting Emile Biasini (the boss on site) to take a taxi and visit La Villette sometime during the works.

Since Rice was engaged with the Arup team, the RFR consulting work for the main pyramid (remember, under the hat of the facade contractor) fell on Martin Francis. Martin was an architect but also a yatch designer. The whole concept of suspended glazing and stainless steel cables owes in fact a lot to the yatching world of mast connections. As part of his sailing activities Martin knew of an American company who were masters in stainless steel rods, cables and riggers for naval architecture. The company name was Navtec and Martin Francis' colleague there was Tim Eliassen. This becomes interesting, and a prove that mixing technologies is always productive. Eliassen, who had studied aeronautical engineering and graduated in nuclear reactors had cofounded Navtec to end up immersed in the world of large sailing yachts, America's cup boats and the like.
Inner transparency, so much sought after by the architects, depends largerly on the point of view.

According to Mic Patterson (himself a good friend of Tim's and as such a reliable source), in 1987 Elliassen received a call from Francis telling him that there was a project in France that needed his involvement. The main pyramid became the first architectural project for Navtec, where they provided about 3,800 'short pieces of yatch rigging' to Eiffel. After completing the Louvre pyramid Eliassen tried - unsuccessfully - to convince his colleagues of Navtec to enter the business of glass architecture. Navtec comment at the time, according to Patterson, was unforgettable: "roofs leak, you get sued". So Eliassen founded TriPyramid Structures - notice the relationship between the name and his first job - in late 1989 and started a long line of high-profile projects, helping US architects to master in the new science of glass and steel. But that's another story.
The complexity of connecting elements next to the bottom of the pyramid. Notice the air fans pointing towards the inner face of glass, intended to reduce the risk of condensation.

A side note about glass selection for the main pyramid will give perspective on how much things have changed since the mid '80s in glass technology. Pei wanted a glass as clear as possible, and he was sure that the required laminated thickness (and even more in diagonal views) would be seen as green form the outside. So under his pressure a new manufacturing process was devised using Fointanebleau white sand -that is, sand very low in iron content - in collaboration with the French firm Saint Gobain. Possible it was, but expensive: the cost of producing a small batch of low-iron glass was huge at the time, and the big boss at Saint Gobain wasn't willing to stop the furnace and introduce such an order. Pei has declared (it's in Jodidio's book) that he went to Mitterrand in person in order to get the glass he wanted. Those were the days: the Emperor stopped the furnaces and glass for the pyramid came out as clear as it had to be...
Section of the concrete slab below the glass pyramid. This was the kingdom of Nicolet Chartrand Knoll, the Canadian engineers who started designing the pyramid concept.

So the main pyramid ended as best as it could, still not as 'trasparent' as Pei wanted, but it would have a great influence - much larger than the serres of La Villette - in expanding the word of the new structural glass facades to the world.

Now, Martin Francis had shown Pei how useful it had been to have RFR on board for the main pyramid, and Pei awarded them with the contract for the inverted pyramid. For RFR the next logical step in the linear sequence of the history of glass had to be the disappearance of mullion frames, elevating glass to the primary structural element of the builiding's skin. And this is exactly what happened with the inverted pyramid: the flat, clean glazed surfaces of La Villette lost their ugly steel tube edges and became a pure glass-enclosed volume. The inverted pyramid is to structural glass what the Seagram building was to the history of curtain walling: the culmination of a de-materialization process that had taken years to achieve. And, Pei permitting, the main pyramid with all its glamour would be nothing but the Lever House, located - as its sibling - just some feet away from the real jewell...

Enough for this post. It's too much text already. Let me finish with some good images and drawings of the inverted pyramid. From the heights of our age it's easy to read and to understand how it works. It surely must have been painful to design, but there it will remail, the light of a candle, forever.
Inverted pyramid as seen from the Carousel

Inverted pyramid: section and structural diagram

Inverted pyramid from below

Inverted pyramid: more structural diagrams

I.M. Pei at the bottom of the inverted pyramid

Inverted pyramid: the square brackets support the top square glass units while the cross brackets support the sloped rhomboid glass units. 
Me as Peter Rice, hanging from cables...

Detail of the top glass brackets. Each glass unit is glued to one side of the bracket at each corner. Slope is 4º for water drainage.
Detail of the side glass brackets. Each glass unit is drilled at the corners. The bracket is a mirror piece to avoid swinging and provide additional stiffening.