Control of computational modelling – Mick Green – presented Aston University 2002
M G Green
It is probably true to say that fire safety engineering is where structural engineering was a 100 years ago except technology and computers are enabling it to progress much more quickly. The technology will support the gradual transition from a prescriptive based approach to a engineered approach where goals can be set and achieved in the style of the more established engineering disciplines. However there is a need for careful planning, validation and a collective approach by all concerned if this transition is to be successful.
1 Computational Models
Computational modelling is a fundamental part of fire safety engineering and will continue to grow as research advances and its application becomes viable. In the aeronautical and petrochemical industries analytical tools are aligned with CAD models to give a much more integrated process. However there is still considerable progress to be made in the construction industry as a whole. Fire safety engineering has to cope with not only the growing pains as a discipline but also the impact of computer power and the ongoing change process. There is no doubt that computational modelling is going to develop so fire safety engineers, regulators and enforcers have to be able to manage it in an efficient, reliable and consistent manner. Confidence in all of the following is essential.
- The people
- The hardware
- The software
- The process
The process depends on the scale of the problem and the level of reliance the engineer is placing on the outcome of the model. The approach described here is typical but there are a few fundamentals that need to be addressed: –
- Has the correct conceptual model been identified?
- Is the way the program is used appropriate and suitable?
- Has the model been created to be a reasonable representation of reality consistent with the intended purpose?
- Are the results, their interpretation and the application correct?
(See use of computers for engineering calculations, The Institution of Structural Engineers)
People are the fundamental part of the modelling process. The roles and numbers of the people will depend on the scale of the project its complexity and the risks involved.
From a designers perspective there is a need to have a sufficient set of skills to include.
- Technicalmanager–charteredengineerorequivalentthathassufficientskilland expertise to take responsibility for the overall project. The primary initial task is to agree the simplest conceptual model that will meet the needs of the design. The conceptual model is principally the decision on whether a 2D, 3D, finite element, CFD etc will meet the needs of the project. In some circumstances a software specialist will be required to enable the selection of the appropriate software. The designer and the enforcer need to be confident that the software developer / academic community has carried out the relevant degree of validation for the intended usage. It is important to recognize that the validation of a model is never complete and is often specific to a particular use.
- Analyst–anexpertintheuseofthesoftware,whohasexpertiseinthecreationofa model so that it reasonably reflects the physical situation to the extent that is necessary for the particular purpose.
- Thecheckerisoftenthetechnicalmanagerbutisatleastsomeonewhoisableto envisage the broader picture and assess compliance with the acceptance criteria. In some circumstances an independent checker would be of value. The primary objective is to verify the results by checking the conceptual model, the input and output. However most importantly taking an over view of the reasonableness of the calculations, perhaps by making simpler global checks is essential.
For smaller simpler projects many of these skills can reside in the same person.
From an enforcers perspective many levels of checking can be implemented from self- certification to a full independent check. Unless there is to be a burden on industry it essential that the designers and the approving authorities maintain the requisite skills to be able to carry out the work, ask the right questions and make the right decisions broadly in line with the process defined below. To help this process there needs to be documents and guidance that all parties subscribe to so that every analysis starts from a well-established foundation of accepted knowledge. Many initiatives are underway but the main challenge is to make sure that there is an effective strategic overview for now and into the future.
3 Software and hardware
ISO 9126 gives some guidance on the appraisal of the six quality characteristics that are relevant in order of importance for software.
- Functionality – is it the right software 2 Reliability – is the software right
Making sure the right hardware is available is fundamental to the process sounds fairly obvious. However if the hardware is not sufficient for the task there is a risk that short cuts and convergence problems will result.
The best processors are very simple once set down. In fire safety engineering there is sometimes a greater intensity of checking than in many engineering disciplines, which has to be recognized in the process that is adopted. There is therefore a need to better integrate the design and the approval process to eliminate abortive effort and poor communication. Some enforcers are satisfied when the designer has suitably skilled engineers involved and that they have adopted a suitable internal review and checking process. However when there is more engagement a simplification of the key stages are as follows: –
- Designers to define the purpose of the modelling, plan the process and select the appropriate people.
- Designer and the enforcer to agree the process that is to be adopted.
- Designer and enforcer to validate the choice of software for the intended purpose 2 Enforcer to agree the conceptual model and the related acceptance criteria prior to carrying out the analysis.
- Designer to carry out the analysis, verify the model and carry out internal checking and review to make sure that the model is sufficient for the intended purpose. Separate sample calculations to check that the answers are of the right order is a very good discipline.
- Enforcer to check that the model complies with the agreed acceptance criteria.
No design is a one way process and so iterations are possible but nevertheless the above should form a basis of the routine. To carry out the whole analysis without progressive agreement is particularly wasteful in time and effort and should clearly be avoided.
5 CFD analysis of the Millennium Dome
The Millennium Dome is located in Greenwich in SE London and has an area of 20 acres, a diameter of 320m and is equivalent to two Wembley stadiums. It is a landmark building on the south bank of the Thames, has a translucent cover and was designed to operate throughout the year 2000 as part of the UK millennium celebrations. One of the primary risks was smoke affecting the escape from the exhibitions and subsequently to the outside on the internal street network.
It was agreed with the local authority that the only way to better understand the implications of a fire inside the Dome was to carry out a CFD analysis. Preliminary Zone model analysis was carried out as an approximate first step in order to understand the scale of problem and to be better able to create the conceptual model and input for the CFD. The purpose of the analysis was to: –
- Help establish the maximum acceptable fire size in the Dome to enable a mini Approved Document B to be written to control the eventual design and construction of the content, which was unknown at the time of the design of the enclosure.
- To help understand the spread of smoke to check that there was a sufficient margin of safety for people to escape from the Dome as part of a time based approach.
Technical issues to be resolved included: –
- Use of CFX software
- It was a complex space with many potential fire scenarios.
- The need to account for the micro – climate, which meant that smoke spread analysis, was required at different times of the year and account had to be taken of various energy sources
- Account of wind effects
- Use of natural and powered extract
- The impact of growing fires
- Fires at different locations
- Layout of the volumetric computational grid 2 Acceptance criteria
- Method of presenting the results
Acceptance criteria for the CFD
Very onerous acceptance criteria were set so that there was a relatively large margin of safety so that every small variation resulting from the above list could be handled in a simple manner. The acceptance criteria were built into method of presenting the results, which was a 50m-visibility iso-surface to represent the approximate base of the smoke layer.
The margin of safety can influence the extent of the analysis necessary for a particular situation. If the margins of safety had been small there would have been a greater need for more refined analysis and more scenarios. However the relatively large margins eventually reduced the scale of the modelling necessary.
6 Evacuation modelling of the Millennium Dome
Evacuation modelling is not a precise technology and involves a significant number of variables and clearly the uncertainties of human behaviour are a major factor. However even though it is not a precise science it adds considerably more to the understanding and the risk assessment process than the simple flow rate scenarios built into the majority of prescriptive codes. The sorts of things that can be accounted for more effectively include.
- Simple blockages in the flow path
- Converging flows
- Opposing flows
- The impact of queues
- Variation in walking speed
More sophisticated models are being developed continuously. However care is need with their use to make sure the conceptual model is correct and that there is sufficient variation in the scenarios considered in order establish the sensitivity of the design to input variations.
7 Structural modelling of GSK House
GSK House is the new United Kingdom headquarters for the Glaxo SmithKline company. The complex in London consists of four buildings, the tallest of which is 15 storeys. All buildings are composite steel-framed structures and have a glazed curtain walling façade. Due to their height, two of the buildings were given a 2-hour fire rating. It was initially proposed to use a board fire protection system to protect all steel beams. However, attaching the boards at the perimeter of the building is difficult. There are also particular considerations with regards to CDM regulations in high rise buildings. Therefore, it was proposed that the perimeter beams be protected with an off-site intumescent. The potential to improve the design and construction process and the health and safety during construction was the challenge for this building.
The detailed target requirements were as follows: –
- To reduce the construction work on site and speed up the construction process
- To reduce the temporary protection necessary at the perimeter of these high rise buildings by increasing offsite production to reduce manual work on site
- Reduction of waste material to be removed from site
The potential solution that would deliver this was the use of offsite intumescent paint on the perimeter beams of the building
Using a prescriptive thickness to achieve a 2-hour rating would have been prohibitively expensive and therefore a model of the real structure in combination with the thermal data derived from tests created the potential for a solution. A number of techniques were used to calculate the thickness, each with its own associated margin of safety. A final thickness was selected such that no one assumption was relied on too heavily and an appropriate margin of safety was selected jointly with building control. This gave a robust solution that all could be confident with.
The factors that were taken into account in the model were: –
- The additional bar reinforcement in the top of the concrete slab used to resist wind shear as part of a verendeel frame.
- Full account was taken of the full 3D-frame action, expansion forces and the interaction between the steel and the composite slab in extreme fire conditions.
8 The typical acceptance criteria for structural fire analysis
There is a general understanding that standards for fire performance of structures, which are based on many real fires and tests, are reasonable and sufficient to meet the broader requirements to limit fire spread, enable means of escape, and allow fire fighting and search and rescue. However there is no real understanding of the level safety provided although it is clearly very varied. Testing and compliance with the following well-known criteria have supported the standard: –
However the use of a more sophisticated approach requires additional consideration in relation to acceptance criteria and in particular deflection. In many cases there will be no need to consider deflection because escape happens at a relatively early stage in the fire development. However there are cases where there may be a need to control deflection. Examples include: –
- Where there is a need to maintain business continuity in a separate compartment or floor
- For fire fighting purposes, where there is a special requirement over and above the normal operations within a building e.g. complex multi use buildings
- Where slabs support compartment walls that could be affected by the same fire that is causing the slab to deflect. The objective is to prevent integrity failure at the junction of a slab and wall
- Where slabs deflect onto a compartment wall due to a fire on one side
- In locations, particularly in large buildings, where there is need to traverse a slab much later in the fire development when deflections could be larger. This could result from phased evacuation or from crossing a slab at the foot of a stair in a high rise building. Escape for disabled people will need to be considered in this respect
- In cases where added fire protection has only been tested up to a certain curvature during the fire test. The risk is that higher curvatures could result in damage to the fire protection unless additional evidence to the contrary can be provided
- Unless a check is made against the above performance criteria there is chance of the failure of a non-structural component as a result of a large deflection even though there is no structural failure.
Advanced engineering methods and analysis methods offer an excellent opportunity to: –
- achieve more consistent levels of safety
- deliver good value
- select the standard of safety, which will be sometimes higher than the prescriptive requirements
- be better able to adopt a risk based approach in major safety critical scenarios