Sunteți pe pagina 1din 18

Construction Innovation 2004; 4: 193–210

Developing a framework for a standardized works


programme for building projects
Karl Blyth and John Lewis School of Architecture and Building Engineering, University of Liverpool,
Liverpool, UK and Ammar Kaka School of the Built Environment, Heriot-Watt University, Edinburgh, UK

Submitted 23 April 2002; accepted 29 June 2004

Abstract: This paper reports on the development of a framework for a standardized


programme of works for construction projects. A sample of 50 buildings, encompassing a
total of 11 different project functions, were surveyed and analysed. The sample was then
investigated further to assess the existence of similarities and repeated operations in each
individual construction project. A minimum of 20 standardized elemental options were
identified. From analysis of the data and the application of practitioners’ expertise, a logical
sequence of activities, including their respective dependencies, was produced. A set of six
new test projects was used to see if the initial methodology was sound. It could be
concluded that despite each project being unique, most buildings retain cognate, elemental
options that provide the basis for any structure, and these can be standardized and used as
a basis for a universal programme of construction works. The standardization of activities
would enable the automation of project planning processes and hence would result in
reduced administration and management costs. This will encourage contractors and other
project team members to undertake planning at earlier stages of the project, hence
providing the basis for more accurate cash flow, duration and cost forecasts.

Key words: elemental design; programme of works; sequencing; standardization;


stereotypes

Introduction

The prime aim of the work described in this paper was to determine if a generic programme of
works could be produced for building projects, at a sufficient level of detail that would allow
for the automation of construction scheduling at the early stages of project planning. The
reduced administrative cost associated with automating the scheduling process would
encourage contractors (and indeed other project team members – traditionally, only contrac-
tors undertake construction planning) to undertake planning at an earlier stage (tendering or
pretendering stage) and hence other associated (where planning should form the basis for
their formation) project management functions, such as cash flow forecasting, resources
allocation, duration and cost estimates could be more accurately generated.
A secondary aim for this element of the project was to investigate the value and role of such
a generic programme in novel developments in UK construction procurement such as those
promoted in recent reports (Egan, 1998) and inherent in the principles of holistic approaches
to procurement such as ‘lean’ and ‘concurrent’ construction and the Private Finance Initiative.

Address for correspondence: Karl Blyth, 22 Regina Road, Walton, Liverpool, L9 2DA, UK. E-mail: karl@
jasonvfreddy.fsnet.co.uk

# Arnold 2004 10.1191=1471417504ci077oa


194 Karl Blyth, John Lewis and Ammar Kaka

In line with the above, the objectives of the work described in this paper are to determine
whether buildings retain cognate, basic, elemental options that provide the basis for any
structure, and to standardize them for use as the basis for a universal programme of
construction works. A set sequence of construction activities would need to be established,
along with their relevant dependencies, to determine start and end dates. A proposed strategy
for calculating the cost, duration, and time lags of these activities is also required. The
timescale and scope of the study is from site set-up to the completion and handover of a
building to its owner.
In order to consider the results of this work in an appropriate context, it is also necessary to
consider the other parallel and potential changes in the construction process as practised in
the UK that might together lead to the increases in productivity and quality of the magnitude
envisaged in the Egan Report (Egan, 1998).

Previous research on the development of generic frameworks


and programmes

This paper is concerned with investigating the frequency with which different design options
(elemental specifications) occur for any given type of building, and the extent to which
designs might be regarded as stereotyped. The products of design can be thought of as various
combinations of known categories (variables) of components (design options).
Harper (1978) and Mackinder and Marvin (1982) all suggested that there was a tendency to
recall and reconsider the familiar when specifying a design. Simon (1975) and Maver (1979)
stated that the effect of stereotyping had the effect of stylizing design, rather than encouraging
innovation. Harper (1978) again suggested that there was a tendency for some clients and
designers to be biased towards a particular design solution, and hence only one design would
be considered. Atkin (1993) stated that it appeared that existing buildings were used far more
as reference points for new buildings than was probably realized. It was further reported that
there should not be any reason why a range of potential design solutions could not be
modelled for a given type of building.
In considering the programming of works, Gray and Little (1985) showed that there is
a commonality of activities for the construction of a building. These activities can be
categorized into major work groups to form an outline construction programme. Nkado
(1992) used these categories as a basis for the development of the construction time
information model. The six categories were set-up, substructure, superstructure, cladding,
services, and finishes. He concluded that site set-up and external works were not critical in
determining overall time. Gray and Little (1985) also found that practitioners use a general
rule to commence a sequence, with the largest locations and those of greatest complexity first,
and concluded that by beginning with the construction of complex locations, the more risky
work will be carried out first. Kahkonen (1992) researched into the development of building
activity sequences, relationships and dependencies and summarized some factors that affect
the sequencing of locations, activities, and overlapping. This can be seen in Table 1. He also
concluded that the start time of the succeeding activity depends upon: 1) the volume of work
space left free while the work of the preceding activity proceeds; and 2) the physical
characteristics of the materials that are drying or hardening.
Framework for a standardized works programme 195
Table 1 Decisions of a planner within schedule preparation and typical factors that have an effect on
these decisions (Kahkonen, 1992)

Decision Typical affecting factor


Sequence of project locations Contractual, Site conditions, Working practice
Sequence of individual activities Structural, Safety, Production technology, Site conditions
Overlap of individual activities Resource, Work area, Safety

Newton (1992) used seven major work categories when analysing high-rise office
buildings: site preparation, substructure, superstructure (frame, staircases, roof, external
walls, internal walls), finishes (wall, floor, ceiling), fitments, services (plumbing, mechanical,
fire, electrical, lifts, special), and site works, of which three of these contained the listed
subcategories. Atkin (1993) proposed that the design of an individual building represents just
one combination of design options that are joined in a predetermined sequence. In addition,
a survey of 40 office buildings was performed in order to study the frequency in which design
options occur. The survey revealed that the frequency with which design options occur
suggested that some were dominant. The highest correspondence between design options for
a pair of buildings was 12 out of a maximum 13 options, for example, frame, stairs, internal
walls, and so on. It was suggested that the existence of stereotypes or themes in design would
be of interest to model builders and users, as it could help them in better defining potential
design solution spaces to problems of design cost evaluation.
Atkin (1993) suggested that designers tended to think in frames. Simon (1975) stated that
restraining influences were likely to lead designers to confine these thoughts within rational
boundaries. The study also stated that Russell (1977), Grogono and Nelson (1982) and
Russell and Powell (1984) all agreed that ideas were rarely visualized in complete isolation
and that the consequence of a simple suggestion might have been for it to be associated
automatically with other ideas.
The above studies all had the same flaw in that there was a low level of internal detail in the
programmes produced. The breakdown of how each standardized activity was established,
how they related to another, how costs and durations were spread, determining relevant
dependencies and the establishment of predecessors were not explained in sufficient detail
and made them unsuitable for some purposes, for example, resource planning and financial
and cost flow prediction. These standardized activities are not suited to this study.

Data collection and analysis

It was necessary to include a wide range of firms in the construction industry to provide a
variety of construction projects. The information was sought from clients, contractors and
quantity surveyors as it was believed that these would be the most likely to yield the detailed
cost and programme information required. The contractors were identified from the NCE
Contractors File 1999 and large retail client organizations were identified via general and
employment publicity material. The 1999 GTI Quantity Surveyors journal (Blythe and Wood,
1999) provided information on the consultants contacted. Information was not sought, for
example, from architects or engineers, as it is believed that they will have not have the access
196 Karl Blyth, John Lewis and Ammar Kaka

to the detailed information on the costs and programmes that are needed. It is not believed
that the scope of the sources will have an impact on the applicability of the results.
The sample consisted of a new set of 200 firms. Each firm was sent the structured multiple-
choice questionnaire, consisting of the 21 questions based upon the characteristics of the
project, along with an accompanying letter explaining the purposes of the data collection,
why they had been asked to provide a project programme with its associated costs, and why
they had been asked to participate. If it was easier for a company to discuss the data
collection, interviews were arranged to collect the data in person, and to discuss any queries
that may have been apparent. The questionnaire was split into two sections. Questions 1–8
consisted of non-numerical (noninterval level data) queries, for example, project type,
location, whereas questions 9–20 were numerically (interval level data) based, for example,
ground floor area, number of floors, and so on. The data required for each project for use in
the model were as follows:
1) The project’s characteristics (based on the multiple-choice questionnaire).
2) The project programme (that is, the bar chart for the construction plan).
3) The individual activity and overall costs (or approximate estimates).
The data were provided by post or obtained personally via interviews. A total of 47 projects
with the above specifications were received. The data on a further 14 projects were obtained
via a continuing series of cost models published in Building Magazine using data provided by
the QS practice Davis, Langdon and Everest (1988–1994). In this, a building was used as an
example each week and a breakdown of the programme of works and the costs of each
activity were provided. Five projects were unfinished and therefore rejected. This made a total
of 56 projects for use in the analysis. This number was considered sufficient to provide a wide
spread of design options (Atkin, 1993).

Determination and identification of the project characteristics


The structure, content and format of the questionnaire were created from a sample in line
with the Nkado (1992) study plus additional questions and also with the principle that
characteristics based around the project size are perceived as being effective in the current
research in that the determination of such information could by obtained quickly and easily by
the respondents in the sample. It was decided to keep the questionnaire as short as possible so
as to ensure a successful response rate, as in the stage payment questionnaire. The question-
naire was split into three sections, reflecting the different areas of data required. Section A
consisted of two basic non-numerical open-ended questions regarding building type and
location. Section B consisted of seven multiple-choice (noninterval level data) queries (apart
from Question 3), whereas Section C consisted of 12 numerical open-ended (interval level
data) queries. A space named ‘Other’ was left after every multiple-choice question in Section
B so as to give each recipient the chance to add an option not already highlighted. Owing to
the huge variability of possible numerical value responses in Section C it made sense to
leave these questions open-ended. It was decided not to include any more characteristics
regarding design options such as foundation type, internal wall type, and so on, as too many
questions would increase the length of the questionnaire and might have left the respondent
uninterested.
Twenty-one qualitative and quantitative project characteristics were measured in the survey.
The breakdown of each can be seen in Table 2.
Framework for a standardized works programme 197
Table 2 The project characteristics

Qualitative data (noninterval level) Quantitative data (interval level)


Building type Construction start
Location Storeys above ground
Type of procurement Height above ground
Main frame Ground floor area
Site access Gross floor area
Service intensity Excavation volume
Presence of atrium Average storey height
Cladding type Volume of building
Ratio of gross floor area to ground floor area
Average floor area per storey
Average volume of storey
Depth of foundations
Ratio of height of building to depth of foundations

Table 3 Breakdown of the project data received

Contractor Quantity surveyor Client Cost models Grand total


Project sample 70 70 60 – 200
Projects received 25 14 8 14 61
Response rate from each (%) 35.71 20.00 13.33 – –
Projects used 24 12 6 14 56
Percentage of overall sample 42.86 21.43 10.71 25 100
Minimum value (£m) 3.47 5.79 3.31 2.20 –
Maximum value (£m) 29.76 15.32 23.44 10.01 –
Total value (£m) 200.68 102.89 51.94 119.97 475.49
Minimum duration (months) 14 11 9 6 –
Maximum duration (months) 33 24 18 12 –

Table 4 Summary of the interval level data project characteristics

Project characteristic Mean Max Median Min Std. deviation


Storeys above ground 2.7 7.0 2.0 1.0 1.6
Height above ground (m) 15.1 35 12.6 5 7.5
Ground floor area (sq. m) 4535.6 21 700 2695.5 31 4908.4
Gross floor area (sq. m) 9868.7 52 500 7000 63 10 083.7
Excavation volume (cub. m) 7255.8 32 550 4884.3 46.5 7891.7
Average storey height (m) 6.5 20 5.2 2.5 3.3
Volume of building (cub. m) 68 315 434 000 48 600 279 84 009.7
Ratio of gross floor area to 2.6 8 2 1 1.8
ground floor area – – – – –
Average floor area per storey (sq. m) 4231.2 20 000 2713.4 31.5 4467.9
Average volume of storey (cub. m) 34115.7 217 000 12 285 139.5 46 726.8
Depth of foundations (m) 1.9 13.9 1.5 0.4 2.4
Ratio of height of building to depth 11.4 35.4 9.2 2 8.1
of foundations
– – – – –
198 Karl Blyth, John Lewis and Ammar Kaka

Characteristics of the data


Table 3 shows the cost and duration breakdown of all of the projects received, including
where they were obtained and their associated response rate. Of the 56 valid projects received,
50 of them were used in the analysis. The six remaining projects were used in testing.
Contractors supplied the most projects, with a total value exceeding £200 million, which also
included the most expensive project. This represented 42% of the total value of all of the
projects analysed. On average these projects tended to take the longest to construct. The cost
models obtained from Building Magazine tended to be small projects in general and took
between six and 12 months. They also yielded the smallest value project to be used in the
analysis. Both the cost models and the projects received from the Quantity Surveyors
represented approximately 25% of the total value of all of the projects.
Table 4 summarizes the characteristics of the interval level data variables for all the projects.
The buildings range from one to seven storeys high, are between 5 and 35 m in height, have
a gross floor area between 63 and 52 500 m2, a volume between 279 and 434 000 m3, an
excavation volume of between 46.5 and 32 550 m3 and foundations ranging from 0.4 to 13.9 m
in height. A summary of the other project characteristics is provided in the Table 5.

Building the framework and predictive model

Standardizing activities and sequences


As all the collected project data differed greatly, especially in the number of activities in each
project, it was necessary to standardize the level of details. The standardization process
involved analysing all of the Gantt charts of the first 50 projects received in order to determine
the level of details and the activities, names that could represent all of these building projects.
First, it was necessary to try to find common design options and related activities. This was
achieved by analysing specific parts of all the programmes of work. This would help
determine whether specific groups of activities would occur at similar times. The activities
were deliberately identified and represented by generic names and hence specific design had
no significant impact on these activities; for example, every building has a substructure,
superstructure, and so on. It became apparent that this was the case in the majority of cases.
The construction phases were mainly in the associated order of set-up, substructure, super-
structure, cladding, services, finishes, and fittings, with only a minimal level of variability. It
was also clear that more than one of the original project activities could be included into
making up a set standardized activity. A total of between one and six subactivities could be
used to produce a single standardized one.
It became clear that the projects could be categorized into three associated groups.
Group A consisted of large open-plan buildings with few internal partitions, such as
distribution centres (warehouses), sports centres, B1 buildings, and libraries. They normally
consist of either one or two storeys, but can have numerous floors. Group B consisted of
social, living quarters with between one and three storeys, medium to large plans, and
medium to high amounts of partitions. These included hotels, student accommodation, social
housing, and so on. Group C consisted of large two or three storey buildings with large plans
and numerous partitions, such as nursing homes, hospitals, and day surgeries.
Each grouping was analysed separately to determine how many activities were common
among the projects. Initial activities in each group were put forward as possible candidates for
Framework for a standardized works programme 199
Table 5 Breakdown summary of the noninterval level data variables

Frequency % of total
PROJECT FUNCTION (PF)
OFFICES 11 22.00
HOUSING 8 16.00
RETAIL 7 14.00
EDUCATION 5 10.00
LEISURE 5 10.00
COMMERCIAL 4 8.00
INDUSTRIAL 4 8.00
GALLERY=MUSEUM 2 4.00
HOSPITAL 2 4.00
SHOPPING CENTRE 1 2.00
AIRPORT TERMINAL 1 2.00
LOCATION (L)
SOUTH EAST 14 28.00
OUTER LONDON 10 20.00
SOUTH WEST 8 16.00
INNER LONDON 6 12.00
MIDLANDS 6 12.00
NORTH WEST 4 8.00
NORTH EAST 1 2.00
SCOTLAND 1 2.00
PROCUREMENT TYPE (PT)
TRADITIONAL 28 56.00
DESIGN AND BUILD 19 38.00
TWO-STAGE TENDERING 2 4.00
CONSTRUCTION MANAGEMENT 1 2.00
MAIN FRAME (MF)
STEEL 31 62.00
CONCRETE 9 18.00
LOAD-BEARING BRICK 7 14.00
TIMBER 2 4.00
MASONRY 1 2.00
SITE ACCESS (SA)
GOOD 20 40.00
AVERAGE 14 28.00
VERY GOOD 10 20.00
BAD 4 8.00
EXCELLENT 2 4.00
VERY BAD 0 0.00
SERVICE INTENSITY (SI)
MEDIUM 20 40.00
HIGH 19 38.00
LOW 11 22.00
ATRIUM PRESENT? (AP)
NO 44 88.00
YES 6 12.00
CLADDING TYPE (CT)
BRICK & BLOCK 13 26.00
BRICK 11 22.00
CURTAIN WALLING 9 18.00
SHEETING RAILS 5 10.00
PORTLAND STONE 4 8.00
BRICK & PORTLAND STONE 3 6.00
STEEL PANELS 2 4.00
BRICK=P. STONE=CURTAIN WALLING 1 2.00
GALVANISED STEEL 1 2.00
HARDWOOD 1 2.00
200 Karl Blyth, John Lewis and Ammar Kaka

standardization. For example, all of the Group A projects had the same following initial
activities: site set-up (planning and mobilization), foundation bases, erection of steel frame,
lift installation, roof coverings, M & E installation, decoration (finishes, floor screeds), and
fixtures, fittings and furniture. Once this was achieved, the next stage was to try and find
commonalities among all of the projects as a whole. It became apparent that all of the projects
had many common activities, especially towards the start and end stages of construction.
A full list was made of common activities that all of the projects shared. This was followed
by the next largest amount of activities that the projects shared, followed by the next largest
and so on, right down to individual unique, activities, such as roof lights. Obviously, as each
project is unique, there are bound to be activities that are specific to each project. This
principle would be an important factor in the process of standardization with respect to the
placing of such unique activities.
After the activities were grouped and standardized, it was necessary to determine a set
sequence. This was primarily based on similarities in the data, but also from a combination of
logic, practitioner’s knowledge of the construction process and with the most effective
sequence for the framework of the modelling in mind. Despite each building being different
in nature, that is, having a different building function, the level of detail in each programme
allowed the authors to determine a set sequence based upon the same substructure principles,
and so on. Although basic similarities were found in activities and sequencing, each building
had differing time frames for each and hence all of the programmes were not the same due to
differing time overlaps.

Amalgamating costs and durations


The costs and duration of each of the original activities had to be analysed and amalgamated
into each new standardized activity. For the cost, it was a case of simply adding the value of
each subactivity included to form the respective standardized total. This was not the case
when standardizing the duration. Table 6 illustrates the possible scenarios encountered when
trying to establish a set duration for each standardized activity. A total of four different
scheduling outcomes were experienced when combining subactivities. For scenario one, the
total duration would be calculated from the start of the first subactivity to the end of the last
subactivity that is, where the activities run continuously. This would also be the case for
scenario three, where a subactivity can begin before its predecessor ends. In scenarios two and
four, the activities do not run continually and hence the total duration is calculated from adding
each of the sections highlighted. Although a single duration is calculated for a standardized
activity, in reality, this constitutes numerous activities starting and finishing at different points
of the standardized time frame. The principles of the previous section also apply.

Proposed modelling for the framework


A number of possible methods were produced. Nkado (1992) surveyed a number of
techniques and decided that multiple linear regression analysis was the most suitable
method to test the objectives of this paper, even for a statistically small sample size (Weisberg,
1980; Gilchrist, 1984; Ireland, 1985).
The characteristics of each project determined via the structured questionnaire were to be
used as the x-variables for the regression. The development of the regression model will be
discussed in detail in other publications. The regression analysis was performed using
a combination of largely automated analysis using SPSS and user-directed analysis using
Framework for a standardized works programme 201
Table 6 Possible scenarios when amalgamating initial activity durations

Scenario Initial activity Bar chart (programme of works)


1) Activities run continuously, as one A
ends, another one starts. There is B
no lag between activities. C

2) Activities are staggered, there is a A


slight delay after the end of one B
activity, until the other begins. C
There are positive lags.

3) Activities begin before their prede- A


cessor ends. There are negative B
lags. C

4) Activities are a combination of A


scenario 2 and 3. There are both B
negative and positive lags present. C

Microsoft Excel.1 Several hundred models were generated, tested and modified or rejected,
using a total of 16 different modelling strategies. Three methodological issues are of interest
in this process: dealing with variability in the input data; dealing with non-numerical
variables, and determining the optimum structure for each model. The final methodology
involved eliminating ‘outliers’, that is, data that truly did not fit the model well.
Not all of the input variables can be expressed directly in numerical form (for example,
location) and cannot therefore be used in a regression analysis without further processing.
Several strategies were tested to deal with the problem. SPSS provides for the problem by
allowing the use of ‘Boolean’ variables. For example, in dealing with building function
each possible value (for example ‘office’, ‘school’, and so on) is assigned a variable that can
be either ‘on’ (value ¼ 1) or ‘off’ (value ¼ 0). Obviously, for any single project only one
variable for building type can be ‘on’. As an alternative, this approach was compared to
the number of patterns of manual allocation of integer values to represent the different values
of a variable (for example, values in the range 1 to 11 in the case of building type). A huge
number of patterns of allocation is possible, but several random patterns and patterns based
on the frequency of occurrence of values were tried. In each case the best method was chosen
by goodness-of-fit to the sample data and the accuracy achieved by the regression models
produced in predicting values for the test projects.
The method of using Boolean variables was initially used based on the integers designated
to the relevant noninterval variable. The designation of the noninterval data can be seen in
Table 7. The choice of service intensity level is now used as an example. There are three
202 Karl Blyth, John Lewis and Ammar Kaka
Table 7 Designated numbers
for the noninterval level data

Designated no.
Site access (SA1–SA6)
1 Excellent
2 Very Good
3 Good
4 Average
5 Bad
6 Very Bad
Service intensity (SI1–SI3)
1 High
2 Medium
3 Low
Main frame (MF1–MF5)
1 Steel
2 Load-bearing Brick
3 Timber
4 Concrete
5 Masonry
Location (L1–L8)
1 Scotland
2 North West
3 North East
4 Midlands
5 South West
6 South East
7 Inner London
8 Outer London
Atrium present (AP1–AP2)
1 No
2 Yes
Procurement type (PT1–PT4)
1 Traditional
2 Design & Build
3 Const. Management
4 Two Stage

possible responses to this variable, namely ‘high’, ‘medium’ and ‘low’. An initial designation
of ‘1’, ‘2’ and ‘3’ was used respectively to represent the aforementioned responses. Figure 1
illustrates some of these conversions with respect to the purposes of correlation and for the
regression.
This was completed in the same way for the rest of the noninterval level data variables. For
the purposes of correlation to determine important variables, as discussed in the next sections,
all columns of converted data must be included, as in Figure 1(b). This is not the case,
however, for the purposes of regression. When a variable has two or more categories as in the
case of service intensity, new variables must be created to represent the categories. The
number of new variables required to represent a categorical variable is one less than
the number of categories. Instead of the three actual values for service intensity, depending
on whether the value was high, medium, or low, two new variables would have to be created
Framework for a standardized works programme 203

Figure 1 The conversion of service intensity

to represent the intensity of services, as illustrated in Figure 1(c). The last column in each case
had to be removed.
As an example, a full set of equations for Activity 2 (see Table 8) – ‘Foundations’ can be
seen as follows:

Duration ¼ 17.285 þ (0.001*project function) þ (0.135*location) þ (1.441*procurement


type) þ (1.217*main frame) þ (1.530*site access) þ (0.837*service intensity) þ
(0.208*atrium present) þ (0.095*cladding type) þ (0.537*storeys above
ground) þ (0.187*height above ground) þ (0003*ground floor area) þ (0.0001*
gross floor area) þ (0.0001*excavation volume) þ (0.038*average storey height) þ
(0.00003*volume of building) þ (2.060*ratio of gross floor area to ground loor area)
Cost ¼ 561 478.43 þ (29 734.810*main frame) þ (60 057.681*site access) þ
(7365.058*service intensity) þ (36 546.358*atrium present) þ (8586.446*cladding
type) þ (52 704.168*storeys above ground) þ (13 945.583*height above ground) þ
(135.373*ground floor area) þ (24.033*gross floor area) þ (53.689*excavation
volume) þ (15 945.478*average storey height) þ (2.925*volume of building þ
(45 671.019*ratio of gross floor area to ground floor area) þ (7.908*average floor area
per storey) þ (3.700*average volume of storey) þ (129 637.347*depth of founda-
tions)
Time lag ¼ 0.162 þ (0.046*project function) þ (0.078*location) þ (0.163*procurement
type) þ (0.034*main frame) þ (0.221*site access) þ (0.009*overall duration) þ
(0.083*service intensity) þ (0.128*atrium present) þ (0.040*cladding type) þ
(0.017*storeys above ground) þ (0.002*height above ground) þ (0.00002*ground
floor area) þ (0.00002*gross floor area) þ (0.00003*excavation volume) þ (0.017*
average storey height) þ (0.000003*volume of building)

Preparation of time lags and dependencies


For the next step of the analysis it was necessary to calculate the time lags from one activity
to another. This would determine when a specific activity would begin. First, each activity had
to be designated a dependent activity previous in the sequence. This is based upon the idea
that an activity later in the construction sequence could not begin until its dependent
predecessor had begun; for example, in practice, watertight would depend upon windows
and external doors, handover and clean would depend upon fixtures and fittings, and so on.
For the purposes of the model, relating to dependent activities in current practice produced
positive lags in many cases, and this was deemed inadequate. For the model to be able to
predict the start times, there had to be no negative lag, that is, the dependent predecessor
204 Karl Blyth, John Lewis and Ammar Kaka
Table 8 The standardized activities in order with their activity dependencies

Activity number Activity name Activity dependency (lags)


1 Site set-up –
2 Foundations Site set-up
3 Drainage Foundations
4 Ground floor Foundations
5 Frame Foundations
6 External walls ground Frame
7 External walls 1st External walls ground
8 External walls 2nd External walls ground
9 External walls 3rd External walls ground
10 External walls 4th External walls ground
11 External walls 5th External walls ground
12 External walls 6th External walls ground
13 Internal walls ground External walls ground
14 Internal walls 1st Internal walls ground
15 Internal walls 2nd Internal walls ground
16 Internal walls 3rd Internal walls ground
17 Internal walls 4th Internal walls ground
18 Internal walls 5th Internal walls ground
19 Internal walls 6th Internal walls ground
20 Internal doors Internal walls ground
21 Lift=stairs 1st floor
22 1st floor Frame
23 2nd floor 1st floor
24 3rd floor 1st floor
25 4th floor 1st floor
26 5th floor 1st floor
27 6th floor 1st floor
28 Roof Frame
29 Watertight (milestone) Internal doors
30 Windows & ext. doors Internal walls
31 Plumbing & sanitary-ware Internal walls
32 Mechanical services Internal walls
33 Electrical services Internal walls
34 Floor finishes Roof
35 Ceiling finishes Internal walls
36 Wall finishes Mechanical Services
37 Fixtures & fittings Mechanical Services
38 External works Foundations
39 Handover & clean Wall finishes

activity had to start before the activity later in the sequence. If the lag was negative, then an
earlier activity in the programme was chosen until there was a zero or positive lag in all of the
projects.
Table 1 does not show the relationships that would be used in producing a construction
programme in practice. They represent the relationships that lead to the most efficient
determination of lag times using regression analysis.
Once this was completed, the next step in calculating the time lags could take place thus:

Absolute time lag ¼ Activity start  Start of predecessor activity (1)


Duration of activity ¼ End date of activity  Start date of activity (2)
Framework for a standardized works programme 205

The time lag expressed as a percentage of the predecessor activity was then calculated.

100  Absolute time lag


Percentage Lag ¼ (3)
Duration of predecessor

These final figures could then be used in the forecasting model. A summary of the results of
the actual versus predicted time lags for the 50 projects can be seen in Table 3. Nkado (1992)
constructed 95% confidence intervals to model the lags, and concluded that the predictions do
not differ greatly from contractors’ estimates.

Costs and duration


All of the standardized costs for each activity were grouped with the rest of the costs for that
activity for all of the projects. Each standardized group of activity costs would act as the
y-variables, to determine the final regression equations. The most effective x-variables,
determined via various methods, would be used to predict the costs in each case. This would
be one for all of the grouped costs. The procedure would then be repeated for the duration.

Utilizing the modelling framework

The final predictive models were constructed on an Excel template, using the regression
equations generated. The template performed any manipulation necessary for preparing
input data for use in the equations and did the necessary repetition of calculation needed to
deal with multistorey buildings of varying sizes. When inputting the maximum number of 21
(coded where necessary in the methods discussed previously) qualitative and quantitative
variables into the template, it immediately produces predictions for individual activity
duration and cost, activity start and end date, as well as the associated time lag from its
chosen predecessor and consolidates these into a single programme prediction.
The predictive data produced were then automatically linked from Microsoft Excel and
transferred into Microsoft Project and presented initially as a Gantt chart. From this various
reports could be produced, for example, the scheduled finish for each activity, delay and float;
the critical path of the project activities; the total duration of the project; and the total project
costs and cumulative monthly=weekly=daily costs. Various reports were also available that
showed; cash flow, budget, earned value, over budget resources and tasks, critical tasks,
milestones, and so on. It was also useful as it could show predecessors, successors and
scheduling constraints for example.

Results and analysis

As all of the collected data differed greatly, especially in the number of activities, it was
necessary to standardize them in order to be able to compare them more accurately. All of the
Gantt charts of the first 50 projects received were analysed to see if any similarities could be
found in their activities.
206 Karl Blyth, John Lewis and Ammar Kaka

The frequency with which specific design options occurred suggested that many could be
regarded as generic. A range of between 20 (minimum, one storey) and 39 (maximum, seven
storeys) activities were developed and standardized for all of the projects. A logical
predetermined sequence of activities, including the respective dependencies based on both
analysing each programme of works and via practitioner’s knowledge of the construction
process, was produced. The results of this can be seen in Table 8.
Table 9 illustrates the different elemental design options encountered during the analysis
that made up the standardized activities. It shows that there are 3 840 000 000 (2  2  2 
2  3  4  4  4  4  5  5  5  5  10  10) potential design solutions. In reality,

Table 9 Design options determined from the 50 projects

Foundations Ground floor Mechanical services


Pad Precast concrete Heating and ventilation
Piles In situ concrete Heating only
Internal walls Electrical services Internal doors
In situ concrete Lighting & power Flush-veneer-hwd. fr.a
Blockwork Lighting=power=comm.c Flush-veneer-swd. fr.b
Metal stud partitions Steel
Panelling
Stairs Upper floors External doors
Precast concrete In situ concrete Galvanized steel
Steel Galvanized steel Aluminium
In situ concrete Precast r.f. concrete Stainless steel
Softwood Timber Glazed=hwd. fr.a
Floor finishes Ceiling finishes Main frame
Carpet tiles Metal pan tiles Steel
Vinyl floor tiles Mineral fibre tiles Load-bearing brick
Ceramic tiles Emulsion paint Timber
Raised access Acoustic tiles Concrete
Granite tiles Plasterboard Masonry
Wall finishes Windows Roof
Plasterboard Aluminium frames Tiles=slates
Ceramic tiles uPVC Metal decking
Laminate wall panels Softwood Concrete=asphalt
Emulsion paint Hardwood Metal cladding
Stone cladding Steel Precast concrete
Built-up felt
Mild steel sheet
External walls (cladding) Galvanized steel panels
Brick & block Aluminium
Brick Single-ply uPVC
Portland stone
Sheeting rails
Curtain walling
Brick=P. stone=C. walling
Galvanized steel
Steel panels
Brick=Portland stone
Hardwood
a
Hardwood frame; bSoftwood frame; cCommunications.
Note: Fixtures & fittings are not shown due to the vast amount of design options for each unique building. This
is the same for External works. Plumbing & sanitary-ware is also not shown due to the options being too
similar in each case, that is, WCs, urinals, basins, and so on. This is the same for drainage.
Framework for a standardized works programme 207

however, this will include infrequently used options and impractical combinations. Hence,
removing design options that represent partial systems or installations, and specific unique
specialist work, would reduce this figure dramatically.
The options listed for mechanical and electrical services can be viewed as degrees of
sophistication (Atkin, 1993). Despite the huge number of potential design options, it was
interesting to note that there was not much variation in the specific activity options. The
variations of each ranged mainly from between two and five different design solutions, with
the two exceptions being the roof and the external cladding types, which both produced a total
of 10 alternatives. There were only two design options for foundations, ground floor, electrical
services, and mechanical services. The standardized drainage and plumbing and sanitary-ware
activities both yielded very similar techniques respectively that could only be categorized into
a single method for each. These results would seem to hint at the existence of clusters of
buildings having similar design options. Every project analysed gave different design options
for both external works, and fixtures and fittings, to suggest the complexity and uniqueness of
these activities.
Table 10 shows a summary of the percentage errors between the actual and predicted data
of the main categories, for the duration, costs, and time lags. It can be seen that there is a
93, 98 and 82% minimum accuracy when predicting these categories, respectively. The
overall costs yielded the most accurate results, and the negligible standard deviation
reinforces the reliability of the proposed model. The overall costs were predicted exactly
on five occasions. The maximum value for the lags is misleading and was produced when
the duration of an activity was very small, that is, less than two weeks. In reality the high
percentage error would only reflect between half and one full week difference from the actual.

Testing the model


To determine the suitability of the activity standardization, the true accuracy of the proposed
regression model, and to further test the results, everything had to be tested on a new set of
projects that were not included in the initial elemental analysis, nor in the development of the
model. Six programmes of work were randomly selected from the initial data set for this
purpose.
The characteristics of the six test projects are illustrated in Table 11. Five different building
types were encountered out of the original 28 building types, to which four of the original 11
project functions were also present. The projects consisted of two offices, two commercial
buildings, one shopping centre, and one industrial project. Five different city locations were
encountered. Three out of the original four procurement types were encountered. The
traditional procurement type was used three times, two-stage tendering twice, and design
and build once. All of the projects had main frames that were steel. Four of the original six
site access options were encountered. Two were very good, an additional two were good, one

Table 10 A summary of the absolute % errors of the actual versus predicted data for the initial 50 projects

Category Mean Minimum Maximum Standard deviation


Overall duration 2.39 0.07 6.47 1.76
Overall cost 0.35 0.00 1.38 0.32
Individual time lags 3.89 1.30 17.45 2.67
208 Karl Blyth, John Lewis and Ammar Kaka
Table 11 Breakdown of the test project characteristics

Project number Cost (£) Duration (weeks) Number of construction


activities
1 22 759 060.93 110.2 25
2 9 056 409.78 53.0 28
3 14 316 226.79 73.0 31
4 5 792 113.26 47.2 19
5 10 006 264.03 54.0 29
6 23 443 839.90 89.0 24

was average and the last one was bad. All three of the service intensity options were
encountered. A high intensity of services occurred four times, followed by once each for
medium and low respectively. None of the buildings had atria. Four of the original ten
cladding types were encountered. Two projects each used brick and block and Portland stone
respectively and once each for sheeting rails and curtain walling.
For the interval level data, the buildings ranged between 1997 and 1999 for commencing
construction, had between one and seven storeys, were between 12 and 24 m in height, had
ground floor areas of between 1500 and 19 877 m2, had gross floor areas between 4500 and
62 507 m2, had excavation volumes of between 2100 and 29 816 m3, average storey heights of
between 3.43 and 12 m, total volumes of between 22 500 and 318 032 m3, had ratios of gross
floor area to ground floor area of between 1 and 6.67, average floor areas of between 1500 and
19 887 m2, average storey volumes of between 7500 and 159 016 m3, foundation depths of
between 1.35 and 2.05 m and had height of building to depth of foundation ratios of between
8.89 and 11.71.
With respect to the design options, the data on all of the original 14 design variables were
collected successfully. No new design options were encountered in the test projects that were
not already present in the initial sample of projects.
The method of activity standardization was successfully applied to the six test projects. The
standardized elemental options were made up from between two and six subactivities. Every
original elemental option present could be placed into one of the standardized activities. No
new design options were produced in each of the test projects. They all consisted of design
options present during the analysis of the initial 50 projects, as can be seen in Table 9. All of
the costs and durations were amalgamated, with no problems. The sequences were also
effective, and the relevant activity dependencies, for use in calculating the time lags, produced
positive lags in every case. Hence there were no problems prior to the proposed regression
analysis.
The regression model was applied to the test projects in order to predict each project’s
overall cost and duration and the lag times between activities. These were then compared with
the actual costs, durations and individual works programmes. It can be seen from Table 12
that the mean accuracy achieved for overall predictions of cost and time was 94 and 96%,
respectively. The accuracy of the detailed predictions of activity time lags was a little lower
and, as was to be expected, given the large number of potential variants involved, showed
significantly greater variation indicated by the values of maximum error and standard
deviation. Also as expected, the test results are less accurate than those for the development
data, but are still considered reliable.
Framework for a standardized works programme 209
Table 12 A summary of the absolute % errors of the actual versus predicted data for the six test projects

Category Mean Minimum Maximum Standard deviation


Overall duration 3.47 0.38 6.68 2.66
Overall cost 5.96 3.71 8.42 1.65
Individual time lags 6.56 4.44 14.52 3.98

Despite the fact that the test projects consisted of four building types (offices, retail,
commercial and industrial) and were unique in themselves, the small values of standard
deviation for all three predictions again suggest that any building in the sample can be further
standardized to provide the basis of a universal programme of works.

Conclusions

The literature review has shown that there is a fundamental limitation of existing theory in this
area of work and that the work that has been done is lacking due to a low level of internal
detail in the programmes produced. For example, the breakdown of how each standardized
activity was established, how they related to another, how costs and durations were spread,
determining relevant dependencies and the establishment of predecessors were not explained
in sufficient detail.
This paper has shown that a group of standardized activities, along with a determined
sequence, and set relevant dependencies can be produced from a sample of 50 UK building
projects. Their designs could be regarded as stereotyped due to the fact that the design options
in the sample are drawn from only a small range of possibilities. The frequency with which
specific design options occurred suggest that some might be dominant. The patterns of these
designs revealed clusters of similar buildings, which indicated the presence of stereotypes.
This paper has also shown that overall project and activity duration, costs, and time lags
can be predicted reliably, via a multiple linear regression model, for an initial sample of 50
UK buildings, based upon a number of noninterval level (qualitative) and interval level
(quantitative) project characteristics. This is confirmed by the accuracy of the results when
tested on a further six projects.
The research has therefore proved that the framework for a standardized programme of
works can be rapidly produced from limited project information, for the buildings in
the sample. The model may be slightly limited due to the small sample size, and the
fact that only 11 building types are encountered. Buildings with special design options would
not be applicable to this study due to the fact that the number of activities would probably
fall outside the maximum number in this research and therefore the model must be used
cautiously for buildings that have characteristics that fall outside those defined and
established in the model. The simplification and standardization of activities, however,
would enable cost, scheduling, and time information models to be built more easily in
the future. This elimination of unnecessary activities would result in increased certainty
in project delivery timescale, reduce administration and management costs, and reduce
project waste.
210 Karl Blyth, John Lewis and Ammar Kaka

It is clear that the framework and associated predictive model will be of significant use at
various points in the procurement process, for example, in client’s and contractor’s estimating.
However, implementing such a proposed framework into the entire procurement process as
currently practised would produce various financial, cultural and social challenges that would
need to be met. For example, should a client organization specify to a contractor that they
should use the client’s construction programme generated from a generic model of works
programmes? This could be seen as robbing the contractor of control of their own operations,
and hence may stifle innovation in construction methods. It would also require changes to
‘standard’ forms of contract and procurement arrangements, and would involve a significant
shift of risk (that is, in telling the contractor how to undertake the construction) back onto the
client, and so on. Although these are significant challenges that need to be addressed, they are
by no means insurmountable, and if the proposed aims of Egan are to be met, then a
compromise must be found in order to benefit all those involved, and to be able to build on the
framework proposed in this paper.
In addition to the work described in this paper, the associated individual and overall activity
duration, costs, and time lags have been investigated and analysed in greater detail. A model
that predicts cash flow for both the traditional interim monthly payment and via a stage
payment approach, that includes the aforementioned model, has also been produced.

References

Atkin, B. 1993: Stereotypes and themes in building Kahkohnen, K.E.E. 1992: Modelling activity
designs: insights for model builders. Construction dependencies for building construction project
Management & Economics 11, 119–30. scheduling. PhD thesis, University of Reading.
Blythe, M. and Wood, A. 1999: GTI quantity Mackinder, M. and Marvin, H. 1982: Design
surveyors journal. The Barns, Oxon, UK: GTI decision-making in architectural practice. Institute
Specialist Publishers. of Advanced Architectural Studies, University of
Davis, Langdon and Everest 1988–94: Building York, Research Paper 19, York, p.114.
Magazine. Maver, T. 1979: The concept of modelling in
Egan, J. 1998: The Egan Report – rethinking architectural design. Design Methods and
construction. London: UK Department of the Theories 13, 67–82.
Environment, Transport and the Regions. Newton, S. 1992: Methods of analysing risk exposure
Gilchrist, W. 1984: Statistical modelling. Chichester: in the cost estimates of high quality offices.
John Wiley and Sons. Construction Management & Economics 10,
Gray, C. and Little, J. 1985: The classification of 431–49.
work packages to determine the relationship Nkado, R.N. 1992: Construction time information
between design and construction. Occasional system for the building industry, Construction
paper No. 18, Department of Construction Management & Economics 10, 489–509.
Management, University of Reading. Russell, B. and Powell, J.A. 1984: Model blindness in
Grogono, P. and Nelson, S.H. 1982: Problem-solving design. Designing for building utilisation London:
& computer programming. London: Addison- Spon, 146–68.
Wesley. Russell, B. 1977: Frames of reference, Design
Harper, D.R. 1978: Building: the process and the Methods and Theories 11, 2–10.
product. Lancaster: Construction Press. Simon, H.A. 1975: Style in design, spatial synthesis in
Ireland, V. 1985: The role of managerial actions in computer-aided building design. London: Applied
the cost, time and quality performance of high Science Publishers.
rise commercial building projects. Construction Weisberg, S. 1980: Applied linear regression.
Management & Economics 3, 59–87. Chichester: John Wiley & Sons.

S-ar putea să vă placă și