Amec is developing this software as part of a larger research project that will make the design process quicker and cheaper. The project's aim is to prioritise design tasks, so cutting the amount of redesign work, reducing overall design time and minimising design-related site problems.
The research began when the company started looking at ways to improve its competitiveness. As a design, construct and management organisation, Amec spends large sums on design fees. It was felt that a lot of effort had been put into rationalising construction on site, but very little into improving the design process. Even with all its designers located in one office, team members tended to focus on design information now within their individual disciplines, for example, the structural engineers concentrated exclusively on the structure, rather than the information requirements of other disciplines.
Connecting the design team
Amec wanted software to link together its design team, but, finding nothing suitable on the market, decided to develop its own. Eight years ago, it joined up with the University of Loughborough. The collaboration resulted in the development of a prototype planning model called the Analytical Design Planning Technique, or ADePT.
The research highlighted the differences between the order in which design tasks are carried out, and the order in which a building is constructed. Simon Austin, a senior lecturer in building and civil engineering at the University of Loughborough, says: "The construction programme forces the building's design to be undertaken out of sequence." For instance, the foundations are completed first, as the structure is built on them. Often, though, to allow work to start on site, the civil engineer is forced to make assumptions about the building's loads because the information is simply not available at that time.
These assumptions, often repeated throughout the project, result in work having to be redesigned once information is finalised.
In an ideal design scenario, says Austin, the foundations would be the last thing to be designed, once all the loads, including plant and equipment weights, were known.
Amec's first step was to create a generic model of the information flow between different members of the design team. Design was expressed in terms of activities and information flows. Although the building itself is often unique, the process model is generic because the design process is the same for all buildings.
To widen the input into the project, Amec brought on board industrial collaborators including Sheppard Robson, Laing, Ove Arup & Partners, Boots The Chemists and BAA. The current model is claimed to apply to 80% of the design process of any new building.
Software used to design computer operating systems was used to model the design process. Over 750 design activities, such as structural loading assessment, and 4000 information transfers, such as air-conditioning plant weight, were fed into the model.
It was applied retrospectively to three case studies. Because the generic model sets out all possible design activities, it must be made "project specific" by deleting all activities not relevant to that project.
The team then set out to logically arrange the order of the design work, based on the availability of design information. The aim was to minimise the number of design assumptions made on a project.
To do this, the team used a matrix, a program that schedules activities in the manufacturing and product design field, and that has been adopted by NASA and Boeing to model their design process.
The output from this matrix defines an optimal order for elements of the design and links elements that are related by an information dependency. For example, the glass specification is linked to the building's heat gain, which then affects the air-conditioning load. This, in turn, affects the size and weight of the air-conditioning plant, and the plant load on the structure.
Finally, ADePT flows the output from the matrix into a project management program. Converting the material into this more familiar format helps the design and construction teams to work out how they can improve their performance. Aspects of the design could be modified to speed up the construction process, for example, load calculations could be moved forward to suit the programme for laying foundations, and aspects of the construction process could then be modified to help the design team, for example, installing false ceilings after all cable sizes had been finalised.
Using project management software also means resources and deadlines can be allocated to tasks, for example, a number of architect days could be allocated to the facade design. In the future, Amec hopes to incorporate this process into the project planning software to further integrate the design and construction process.
"The advantage of this method," says the University of Loughborough's Austin, "is that it takes the risk out of making design assumptions." It presents graphically what has been assumed, where overdesign has taken place because assumptions have been made, and allows the overdesign to be costed.
And the system has a built-in safety mechanism. Once the information is available, it can be fed into the program to check that the assumptions were correct.
Costing design changes
Modelling the design process helps the team gain an understanding of how its decisions affect the project cost, and the visual output emphasises the information linkage of design tasks, highlighting critical areas of information flow. Designers can see how their input impinges on other parts of the team.
Plotting the interdependency of design tasks means the research project can assess the implications of client changes. The knock-on effect of changes to the design can be seen both in redesign work and on the programme timetable, and allows such changes to be costed accurately. Likewise, if a design-and-build contractor requests the early release of information on, say, the structure, the assumptions made in releasing the information early can be clearly seen.
The team is now testing the software on live projects. Amec is set to trial it on a large pharmaceutical project, and collaborator Laing has sent the software up to Scotland to test it on its Glasgow Royal Infirmary project.
"At this stage, it is difficult to put a value on the time and cost savings ADePT will generate," says David Scott of Laing, one of Amec's industrial collaborators in the project.
So far, more than £210 000 has been spent on the work, half from the government.
The project was completed in October 1998 and the results of the research are being disseminated at conferences in 1999.
Amec is now looking to extend the scheme into the concept stages of a project and into procurement, to gain the benefits of supply-chain integration.
The team's next move will be to get a software house to develop the program as a commercial product. First, the team has to convince such a firm that a market exists for the software, which is proving a difficult task.
