The U.S. is not the first country to engage in such an effort--in fact, it is one of the last major industrial countries to start such work. This is not a disadvantage, since the experience gained from examining previous efforts can be applied. Such earlier efforts have been noted in Australia, New Zealand, Canada, the United Kingdom, the Nordic countries (under their NBK umbrella group) and other places. Also, an effort has been going on internationally, under the auspices of the International Organization for Standardization (ISO), whose TC 92/SC4 has been producing documents on this theme.

Work on developing a performance-based approach has gone on in Japan since 1983. Here, we will specifically only examine the fire aspects of developing a performance-based building code. The present author is the official U.S. delegate to the current effort by Japan's Ministry of Construction (MOC) to develop rational bases for fire safety design on performance-based principles. The purpose of this short paper is to examine some highlights of the Japanese activity and to point out how certain lessons learned there could be applied towards the U.S. effort.

In response to the seen limitations of the earlier scheme, in 1993 the Japanese Ministry of Construction commenced a new 5-year project on this topic. The project is called SOPRO, denoting 'Sogo Gijutsu Kaihatsu Project,' which means Comprehensive Project for Technical Development. The project is run by the ministry's Building Research Institute, with the work being done co-operatively by BRI staff and Japanese industry. The present author has participated in this work as the U.S. expert assigned to the supervisory committee of the project.

The SOPRO project has a completion date of 1998. The projected date for implementing a performance-based track into the Japanese building law is the year 2001. During the subsequent three years after the SOPRO project is completed, two activities are planned: (a) technical work is expected to continue in a "follow-on SOPRO," which will be launched in 1999 for either a 3- or 5-year period. This one will be slightly narrower in scope, focusing on fire + earthquake issues. (b) Decision-making on implementation will start within a new Council for the Promotion of New Construction Structure Systems. The latter was constituted in 1996, but has not yet had much to do, since the technical SOPRO work is not complete.

• The replacement of internal Japanese fire test methods by ISO standards.
• The technical improvement of test methods, including new inventions, as needed.
• The development of new performance-based design guidelines which only maintain an overall equivalent level of safety to the current prescriptive system.
• International co-ordination

The first two elements concerning test methods are further subdivided into 3 work groups:

1. reaction-to-fire tests
2. structural fire performance
3. assessment of building mechanical equipment

In this paper, we will focus on the element which comprises the development of new performance-based FSE guidelines.

The general objectives of the FSE component of SOPRO are set out as:

1. Ensuring transparency of requirements and standards
2. Maintaining an overall level of safety equal to that of the previous prescriptive law
3. Introducing performance-based standards
4. Ensuring consistency of standards.

• specified loadings
• specified safety factors
• mandated fire scenarios, and
• quantitative criteria, expressed as equations explicitly describing the pass/fail demarcation

The most striking feature of both of these Western documents is that they are totally non-mandatory and lack quantitative criteria. In the BSI document, it explicitly states that "At an early stage of the design process, the objectives of the fire-safety design should be clearly defined and the acceptance criteria established." In other words, permission is given to the architect to select both his own level of safety and his own criteria for demonstrating he has achieved an adequate design. In practice, one has to presume that if the BSI CoP is adopted, that local building officials will place some limits to this freedom. Nonetheless, it is clear that in the BSI view, neither a national level of safety, nor nationally-applicable criteria are necessary.

The ISO approach is perhaps even vaguer, since the level of safety is nowhere considered, let alone mandated. The designer is basically instructed to design various aspects of fire safety "adequately," with no notion given of how adequacy will be determined or enforced.

In the view of the present author, the enunciating of pass/fail criteria is essential to the functioning of a performance-based building regulation. Once the phenomena have been quantified and numbers are presented, it must become clear whether or not the expected criteria are met. In the BSI/ISO approaches, this is not made explicit and is deferred to some subsequent evaluation process. This philosophy is likely to create significant uncertainty and difficulties in implementation. Essentially, the BSI/ISO approaches only quantify certain features of the fire/building/human interaction. They provide no benchmark against which to judge if the outcome is bad, good or borderline. But if the issue of "what is good enough performance" could not be successfully determined by the assembled experts in the BSI and ISO committees, there does not seem to be much chance that such judgements could be adequately made by officials reviewing building plans.

The SOPRO approach, on the other hand, makes it clear that the criteria must be built in to the system of equations. The officials reviewing the plans have to scrutinize the designs to make sure that they indeed conform to the performance criteria demanded, but they are not presented with the impossible task of formulating performance criteria. (The alternative, of officials simply accepting performance criteria proposed by building architects, seems hardly adequate or equitable).

Compared to the BSI/ISO approach, there are two essential differences, one on a policy level and one on an engineering level. The policy issue is identified in objectives 2 and 4 above: The SOPRO developers insist on a specified level of safety, and this level of safety must be nationally applied. In parallel, Objective 4 ensures that the standards (i.e., criteria) are also nationally applied. Thus, from an engineering viewpoint, the most striking difference between the SOPRO approach and that of BSI/ISO is the way that the developers have approached the technical task. The BSI approach was to subdivide the necessary fire performance into small elements, then to seek published scientific equations for computing the physical phenomena involved in each element. The 'work output' can be seen to be mainly this collection of equations for a multiplicity of fire phenomena. The ISO approach originally was identical, but the final balloted documents lean towards simply providing references to equations, rather than actually listing equations.

Another aspect in which the Japanese work is different from the Western approaches is in its treatment of the research base. In the BSI/ISO proposals (and in many other Western schemes) the developers have only taken on themselves to cull the scientific literature and to cite suitable formulas. Where a phenomenon has been inadequately studied, the Code of Practice simply contains a placeholder. The Japanese philosophy, on the other hand, considers that in such cases a focused, short-term research project must be mounted. This philosophy was already operable during the predecessor Manual of Practice in 1988 and is equally operative during the SOPRO work.

(a) Fire compartmentation
In Western building regulations, the logic behind fire compartmentation tends to be very opaque. It is clear to all that the reason for it is that a limited-size fire may be possible for fire services to control, but beyond a certain size, their capabilities may be overrun. However, the engineering requirements to achieve this generally do not exhibit much logic and are viewed as legal heritage. The situation in Japan has been not too different. With the SOPRO work, at attempt is made to establish compartmentation on a rational basis. One of the main chapters of the document is entitled "Fire safety of compartments," and it encompasses not only room/floor subdivisions, but also atria, multiple buildings in a connected complex, and similar geometries. Unlike in today's approach, the objective here is not to merely satisfy the requirements of a fire resistance test, but rather to rationally provide for limits to heat and smoke entering from one compartment to another or else structural damage propagating to another. A special treatment is given to computing the viability of compartments which are to serve as shelters.

(b) Building occupancy
Today building regulations view occupancies on the basis of whether the building is a hotel, a factory, a school, etc. The SOPRO approach sets out a multi-dimensional classification scheme for building occupancy:

• sleeping: yes/no
• main population: able-bodied or not
• familiarity: are most persons in building intimate with its layout?
• density of people
• fuel loading
• combustibility of fuel load
• hazard of ignition occurring

(c) Firefighting access
The Japanese, even in the current regulations, have quantified more aspects of firefighting access than is normally seen in the West. In the new SOPRO Code, for example, there is an explicit classification of roads around buildings into ones which are needed for firefighting access, versus ones which are not. For the roads which are needed for access, special structural provisions are made to ensure that the building will not collapse onto the road for the whole duration of the fire.

P Performance-based criteria

C Complementary criteria [=simplified performance criteria]

D Deemed-to-satisfy criteria [=tentative criteria, to be worked upon later]

S Specification [= prescriptive] criteria

E Expert-opinion criteria

The P criteria will encompass the fully performance-based aspects. The C criteria are either partly-empirical computation methods, or else simplified criteria for use in smaller buildings, where extensive calculations may not be justified. The categories D and E are interim measures, where it is recognized that a need exists for a more performance-based approach, but such a one is not ready yet. A certain amount of S criteria will remain for practical reasons.

Examples of each may be given:

P criteria. The heat to which the escaping occupants are being subjected on the exit path is expressed as a computational formula, to be determined from an analysis of fire conditions, walls, leakage from openings, etc. 

C criteria. Requirements are prescribed for areas of refuge which are located in enclosed open-air courtyards and the like. One of the requirements is a formula which relates the distance the refuge must be from the building, vs. the building height. This is to prevent evacuees from being hit by falling debris from the burning building. 

S criteria. The dimensional requirements for row spacing of theatre seats is given as an S criterion. This is because such numerical values have already been proven out in use, and there is no reasonable basis for proposing a varying engineering solution.

E criteria. There may be no objects which can fall on top of evacuees along the exit path.

• During the course of some investigations at the Building Research Institute, the researchers found that walls with grooves in them tend to have quite different fire propagation characteristics from flat walls. Such walls are often specified by architects in auditoriums. The final SOPRO report is to contain special provisions for the testing and analysis of such wall surfaces, based on equations being developed for this case.
  
  
• The use of intermediate-scale heat release rate test equipment is just starting to be considered in ISO. Such work has been going on in Japan for a number of years, however, and criteria based on it are expected to be included in the final SOPRO report. For the majority of products, small scale testing according to the small-scale ISO 5660 Cone Calorimeter will be entirely adequate. For validation work and for highly unusual structures, large scale testing according to the ISO 9705 room test is necessary, and this test is being considered by the European Commission as the benchmark for fire propagation assessment. There are products, however, whose irregularities can make them unsuitable for small scale testing, yet which do not present novel performance issues requiring large scale testing. The Japanese researchers have been aware of this issue for a number of years, have built intermediate scale apparatuses, and are now developing relevant performance criteria.
  
  
• A procedure for medium-temperature testing of fire doors exists in draft form in ISO (CD 5925-1). Very few European laboratories have implemented this test. The Japanese researchers investigated the test and the suggested construction drawings for the test furnace. They concluded that the design is not viable and launched a small project to develop a practical furnace design which could achieve the objectives of the test. This has now been completed and a number of tests have been made with the equipment. It is their intention to feed this information back to ISO.

• a uniform minimum level of safety, independent of who designed the building or which official reviewed the plans.
  
  
• specified loadings, safety factors, fire scenarios (as a minimum; the architect, of course, remains free to design a higher level of safety if asked for by his client).
  
  
• explicit, quantitative evaluation criteria.