Examination of relationships between building form ...

This article was downloaded by: [Nottingham Trent University] On: 27 May 2015, At: 07:00 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Construction Management and Economics Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rcme20

Examination of relationships between building form and function, and the cost of mechanical and electrical services L. M. SWAFFIELD & C. L. PASQUIRE Published online: 21 Oct 2010.

To cite this article: L. M. SWAFFIELD & C. L. PASQUIRE (1999) Examination of relationships between building form and function, and the cost of mechanical and electrical services, Construction Management and Economics, 17:4, 483-492, DOI: 10.1080/014461999371402 To link to this article: http://dx.doi.org/10.1080/014461999371402

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Construction Management and Economics (1999) 17, 483± 492

Examination of relationships between building form and function, and the cost of mechanical and electrical services L.M. SWAFFIELD and C.L. PASQUIRE Department of Civil and Building Engineering, Loughborough University, Leicestershire LE11 3TU, UK

Downloaded by [Nottingham Trent University] at 07:00 27 May 2015

Received 22 September 1997; accepted 22 January 1998

This paper describes analysis work undertaken to examine relationships between building function, building form and mechanical and electrical services cost, including the collection of raw data, and the transformation work undertaken to enable analysis. Relationships are identi® ed between building form parameters, e.g. perimeter of external walls, gross ¯ oor area, storey heights, percentage of glazing, and the mechanical and electrical services costs for buildings of different functions (commercial, industrial and residential). There are relationships between the costs of the mechanical and electrical services installations and some building form descriptors, but the particular descriptors and the strength of the relationships vary according to the function of the building. Keywords: Mechanical and electrical services, tender cost, cost planning, building function, building form

Introduction Under traditional methods of budget estimating and pre-contract cost planning, cost estimates for mechanical and electrical (M&E) services generally were based solely on the gross ¯ oor area of a building. Swaf® eld and Pasquire (1995) argued that it was inappropriate to use gross ¯ oor area as the sole descriptor for determining M&E services cost. Swaf® eld and Pasquire (1996) attempted to predict M&E services costs using gross ¯ oor area as one of several independent variables (using data published by the Building Cost Information Service (BCIS) on detailed cost analyses), and found that there was not a strong relationship between M&E services cost and the building form descriptors used by the BCIS. Swaf® eld and Pasquire (1996) also argued that building form descriptors other than those published in the BCIS detailed cost analyses, may be useful for estimating M&E services cost. This paper describes the selection of building form descriptors appropriate for the research (including the development of new 0144± 6193 1998 E & FN Spon

descriptors), the collection of data for the building descriptors, the transformation work and analysis undertaken, and the results, taking building function into consideration. Figure 1 summarizes the analysis work undertaken and demonstrates how M&E services cost was related to building form, building function, M&E services performance (numerical expression of designed provision, such as boiler rating in kilowatts, speed and weight capacity of lifts, or illumination level in lux) and M&E services quality (material speci® cations determined by aesthetic considerations, such as lift car ® nishes, types of luminaire or heat emitter).

Building function Brown (1987) identi® ed a relationship between building function and M&E services cost. From a sample containing factories, of® ces, housing, general hospitals, health centres, sports halls, primary schools and sheltered housing, Brown (1987) established that


484 the `cost of building services’ to `cost of scheme’ ratios came from different populations, and therefore the percentage cost of M&E services varied between building functions. Brown (1987) found also that the allocation of the total services cost among the various services elements could not be predicted accurately from knowledge of building function. Therefore, this research used the total M&E services cost for the analysis, rather than more detailed breakdowns of the M&E services cost among the elements. It was argued that the function of the building affected M&E services cost because building function determined M&E services performance and quality, and could affect the design of the systems. For example, McLellan (1995) described how the horizontal distribution of M&E services varied between the different laboratory facilities in Glaxo’ s medical research centre, because of the maintenance requirements that arose from the operations carried out in each type of laboratory. Leivers (1995) believed that the function of the building affected the relationship between source plant capacity and cost, because certain buildings (such as hospitals) required back-up plant in case of breakdowns or routine servicing/maintenance. Statutory regulations played a large part in determining the quality of terminal outlets for particular types of buildings, which also affected cost (Leivers, 1995).

Building form parameters Kouskoulas and Koehn (1974) argued that the cost of a building was a function of many variables, and a set of independent variables should be selected that described a project and de® ned its cost. Such variables must be measurable for each new building project. Kouskoulas and Koehn identi® ed the following independent variables that de® ned the cost of a building: building locality, price index, building type, building height, building quality, and building technology. Their cost estimation function involved assessing values for each of the identi® ed variables, and was therefore rather subjective. Fletcher (1968) believed that the identi® cation and proving of useful cost parameters for M&E services was hindered by the prevalence of contracts based on drawings and speci® cation, and the associated absence of detailed analyses of the M&E services cost. In order to test the hypothesis that building form was related to M&E services cost, it was necessary to identify variables that described building form accurately. Brandon (1978) identi® ed the following as suitable descriptors of building form: plan shape index, number of storeys, boundary coef® cient, average storey height, percentage of glazed area, and plan compact-

Swaf® eld and Pasquire

Figure 1 Diagram of relationship analysed

ness. Swaf® eld and Pasquire (1996) identi® ed percentage of glazed wall area, perimeter length, total building height, volume of plant rooms and services cores, and volume of air handled by HVAC systems, as descriptors that may be useful for determining M&E services cost.

Identi® cation of data required The information published by the BCIS had been inadequate for forecasting M&E services cost because of the parameters used for collecting information and the lack of detail in the information submitted by subscribers (Swaf® eld and Pasquire, 1996). Therefore further information had to be collected that was suitable for the necessary analysis work. Raw project cost information from previous projects was required, together with building form and function details, and speci® cation data concerning the M&E services contained in the buildings.

Development of building form descriptors Swaf® eld and Pasquire (1996) did not establish whether the poor relationships observed between M&E services cost and the information published by the BCIS, were due to the descriptors used or the way that subscribers analysed the costs of previous projects. Therefore it was decided that the BCIS descriptors would be included in this analysis, to identify their suitability for determining M&E services costs when all the ® gures were analysed in a consistent manner. Some new building form descriptors were developed for the analysis (Table 1) after M&E services design literature had been studied to identify factors that were relevant to the design of various M&E systems. These factors were then examined in relation to project drawings, and building form parameters already in use. The new descriptors represented building form parameters that were relevant to M&E systems design, and could

Mechanical and electrical services

485

Table 1 Building form descriptors developed for the research Descriptor

De® nition


Floor to ¯ oor height

Height measured from top of structural ¯ oor to underside of structural ¯ oor Total height Sum of all ¯ oor to ¯ oor heights Usable height Sum of usable heights at each storey Usable volume Sum of plan areas of each ¯ oor multiplied by the usable heights at each storey (Note: usable volume = internal cube ± horizontal distribution volume) Plant room area Floor area used for lift, tank and plant rooms, measured on internal structural face of enclosing walls Plant room Plant room area multiplied by usable volume height of plant rooms Horizontal (Total height ± usable height) ´ average distribution plan ¯ oor area volume Vertical Usable height ´ plan area of vertical distribution distribution space (such as service volume cores) Glazed area Area of windows, glazed doors, and panels Internal Perimeter of building measured on perimeter internal structural face of enclosing length walls

be measured from tender drawings, but were not used currently for analysing costs. The building form descriptors as de® ned by the BCIS also were used where applicable, except ancillary area and storey height (see below). Previously published building descriptors also were identi® ed and tested. These were: plan shape index (Banks, 1974), plan compactness parameter or POP ratio (Strathclyde University, cited by Ferry and Brandon, 1991), square index (RICS, 1982), and perimeter index (J. Cooke cited by Ferry and Brandon, 1991). Figure 2 shows some of the building form descriptors used for the analysis.

Figure 2 Building form descriptors : area, height, volume and enclosure

ratios were calculated that represented relationships between some of the building form descriptors (Table 2). These derived variables emphasized relationships with plant room space, and the amount of glazing. The space required for plant rooms was an indication of the intensity of the M&E services provision in the building (Figure 2). For example, a highly serviced building such as a laboratory or a hospital, would require more space for plant rooms than a building with less complex M&E services installations, such as a primary school. The extent of glazing in a building was believed to be signi® cant when considering the M&E services cost because glazed areas affected heat loss, solar gain, lighting and ventilation (if opening windows) requirements (Crane, 1995). Table 2 Ratios calculated from building form descriptors

Relationships between building form descriptors It was proposed that the M&E service requirements of a building had implications on the size of ceiling voids, storey heights, and usable to gross ¯ oor area ratios, because of space required for plant rooms and risers (Chelmick, 1995). To enable more detailed analysis,

Average storey height Wall:¯ oor ratio Percentage glazed wall area (for each elevation and total) Glazing:¯ oor ratio Plant room area as a percentage of gross ¯ oor area Plant room area:usable area ratio Plant room volume:usable volume ratio Plant room volume as a percentage of internal cube

486



Data collection Information containing the building form descriptors identi® ed as worthy of further analysis was not available in a published form. Therefore a method of collecting the required information had to be devised. It was established that information for many of the building form descriptors could be taken from tender drawings. Therefore it was necessary to collect priced tender documents and analyse the M&E services costs against the quantities calculated for the building form descriptors. Due to dif® culties in obtaining data considered commercially sensitive, usable information was obtained for only ® fteen projects. However, the volume of work involved in measuring quantities for the building form descriptors from the tender drawings, and the detailed analysis subsequently required, meant that large sample sizes were not practical anyway. The data collection dif® culty also meant that there was no control over the building functions of the tender information collected. In addition, the sample contained refurbishment and extension projects, which were deemed to be inappropriate to analyse (at this stage) with respect to the building form descriptors. These limitations resulted in the detailed analysis of twelve new build projects: ® ve of® ce buildings, ® ve university student accommodation projects, one document store, and one factory with integral of® ce accommodation. The projects were grouped under broader building functions of commercial (of® ces), residential (student accommodation), and industrial (document warehouse and factory).

Re-de® nition of some descriptors While analysing the tender documents collected it became apparent that some of the descriptors were not entirely appropriate to the data set. It was therefore necessary to re-de® ne some of the descriptors, and others had to be omitted from the analysis (see below). The BCIS de® nition of storey height was the `height measured from ¯ oor ® nish to ¯ oor ® nish’ (BCIS, 1969). Ceiling void depths were affected by the M&E services requirements of the building (Chelmick, 1995), and therefore were an indication of the intensity of the M&E services provision in the building (Figure 2). This research deviated from the BCIS de® nition of storey height, and measured height from ¯ oor ® nish to underside of ceiling ® nish, thus considering only usable height (Figure 3). This enabled the introduction of a new parameter, the `¯ oor to ¯ oor height’ (Table 1). The depth of the service voids (whether suspended ceilings, raised ¯ oors or both) could then be calculated, as the total height minus the usable height, and included in the analysis.

Figure 3 Building height descriptors

BCIS (1969) de® ned ancillary area as the `total area of all enclosed spaces for lavatories, cloakrooms, kitchens, cleaners’ rooms, lift, plant and tank rooms and the like, supplementary to the main function of the building’ . This research was particularly interested in the space required for plant rooms, as an approximation of the intensity of the M&E services provision, (as discussed above). Therefore it was deemed inappropriate to include plant rooms in the ancillary area. Many of the residential building projects analysed had en-suite toilets and shower rooms to each bedroom, and kitchens to share between a group of bedrooms. Therefore it was seen as more appropriate to include these toilets and kitchens with usable area. This resulted in the re-de® nition of the usable area descriptor (to include the BCIS ancillary area items with the exception of lift, plant and tank rooms), the development of a new descriptor for plant room area (Table 1), and the re-measurement of some quantities from the tender drawings, in line with the revised de® nitions of the building form descriptors. Examination of the tender documents obtained for the research revealed that only two of the projects had basements, therefore it was deemed inappropriate to include basement area in the analysis. Two of the projects were largely single storey, and therefore upper

487


Mechanical and electrical services ¯ oor area, and average storey height above ground ¯ oor were inappropriate for the analysis of the available data, and were discounted. To simplify the analysis work, ground ¯ oor area and average storey height at ground ¯ oor also were discounted from the analysis, and storey heights were averaged over all ¯ oors whether above or at ground level. Vertical distribution volume had to be omitted from the research because it was not possible to identify the location (and therefore calculate the total plan area) of the service cores/ducts for all the projects. The drawings for the highly serviced multi-storey of® ce projects showed the locations of cores, dry risers, stair and lobby pressurization, car park and WC extract ducts, etc.; however, the drawings for the residential projects were far less detailed. The student accommodation projects typically showed furniture arrangements in each bedroom (desk, bed, etc.) but contained very little information about vertical services distribution.

Data transformation Due to physical differences between the building projects, and the economic conditions at the tender dates of the various projects (the projects analysed were tendered between December 1991 and June 1996), several adjustments were required to minimize tender price differentials. The tender price of a project was affected by many factors. Smith (1995) found that M&E services tender costs were in¯ uenced by location, programme, terms and conditions of contract, nature of the site and the state of the market. Adjustments were made to the actual tender ® gures using published indices and factors (BCIS, 1997a,b) intended speci® cally for use when comparing costs of different projects. The factors used for the research minimized variations in M&E tender costs due to the different locations, contract sums, procurement routes, building heights, types of work and building functions of the projects analysed. Brief descriptions of the BCIS indices and factors are given in Table 3. BCIS (1977a) described how project factors could be used to show the effect of certain variables on contractors’ pricing. In this research, the range of building forms and functions, locations, time periods, contract sizes and types of building work required meant that many variables could have explained the variation in tender prices between the projects. The published individual adjustment factors for the various physical and economic differences (such as number of storeys, geographical location, etc.) were multiplied together, as described by BCIS (1997a), to produce project factors speci® cally for this analysis.

Table 3 BCIS adjustments made to actual tender ® gures Tender price indices

Market conditions index

Location

Size of contract

Procurement route

Type of work

Building form

Building function

Measured the trend of contractors’ pricing levels in accepted tenders (BCIS, 1997b). The all-in tender price index was selected for the analysis work Based on the tender price index de¯ ated by the general building cost (excluding M&E) index. It was therefore an indication of the competitiveness of the tendering climate The BCIS location factors (BCIS, 1997a) attempt to identify some general building cost differences due to localized variables such as demand and supply of labour and materials, workload, taxation and grants, the physical characteristics of a particular site, its size, accessibility and topography A factor representing the general relationship between price levels, as measured by the tender price index and contract size at 1985 prices (BCIS, 1997a) Adjustment factors relating to the variation in price levels between competitive, negotiated and serial contracts (BCIS, 1997a) Adjustment factors for new build, horizontal extension, vertical extension, shell only or rehabilitation/conversion projects (BCIS, 1997a) The only BCIS adjustment for building form was for the height element, which was represented by the number of storeys factor (BCIS, 1997a) Adjustment factors relating to the variation in tender price levels between buildings of different functions (arranged in broad CI/SfB Table 0 (RIBA, 1969) categories) were published (BCIS, 1997a)

In this study, the main focus was the cost of the M&E services element of the tender, so factors for the size of the M&E services element of the contract were included in the way described by BCIS (1997a) for the size of contract (Table 3). The market conditions index was included in the project factors developed for this analysis as the competitiveness of the tendering climate was believed to be signi® cant in determining the tender prices. Table 3 shows that the BCIS calculations for the market conditions index excluded costs of mechanical, electrical and lift installation works from the general building cost index used to de¯ ate the TPI. Therefore any peculiarities in the competitiveness of the M&E sector would not be accounted for when considering the adjustments to M&E tender costs. The project factors developed for this analysis were as follows.


488


Factor 1

T ´ L ´ N ´ C ´ P ´ W´ B

Factor 2

T ´ L ´ N ´ S´ P ´ W ´ B

Factor 3

M´ L ´ N ´ C ´ P ´ W ´ B

Factor 4

M´ L ´ N ´ S ´ P ´ W´ B

Here, T is the tender price index, M is the market conditions index, L is the location factor, N is the number of storeys factor, C is the size of contract factor based on total contract sum, S is the size of contract factor based on M&E services element of the contract sum, P is the procurement route factor, W is the type of work factor, and B is the building type factor. The market conditions index was derived from the tender price index (Table 3), so these two adjustments were not included in the same project factor. The project factors above included either C or S, but it may have been useful to develop further factors that incorporated both of these variables in the same factor. However, this was not done because it was believed that four project factors would be suf® cient for the analysis at this stage.

Analysis of data The analysis was undertaken in three stages: ® rst, the ® gures were analysed by inspection; then sample means and standard deviations were calculated and examined; ® nally, the data were analysed for correlations between the data sets for the variables used. At each stage the data were analysed as a sample of twelve projects, and then re-grouped by building function.

Inspection When the data were inspected as one sample, the variation in scale and cost of the projects was apparent: the projects ranged in contract value from £871 415 to £25 799 701; in gross ¯ oor area from 1 231 m2 to 24 315 m2; and in M&E services cost from £311 390 to £6 032 706. These variations in cost and size, together with the range of building functions and M&E

services provision, made it unlikely that any signi® cant relationships would be observed between building form and function and M&E services cost for this sample. The residential projects ranged in size from 3 757 m2 to 11 376 m2 gross ¯ oor area, and actual M&E services tender cost ranged from £456 546 to £1 534 008. The smallest project in terms of gross ¯ oor area did not have the lowest M&E services cost, but did have the lowest total contract value. This indicated that there may not be a relationship between gross ¯ oor area and M&E services cost for residential projects. Inspection of the ® gures for residential projects revealed that relationships between M&E services cost and the building form descriptors and ratios appeared to be reasonably signi® cant for four of the projects. The other project (residential 3) was considerably larger than the rest, and appeared to be outside the range of the relationships observed for many of the variables. For example, percentage of glazed wall area showed an interesting relationship with M&E services cost for residential buildings (Table 4). The actual M&E services cost per percentage of glazed wall area was around £50 000 for four of the projects, but over £120 000 for residential 3, even though the percentage of glazing was within the range of the sample. The range was narrowed by the adjustments applied to the M&E services cost data set, with factor 2 producing the smallest spread of values. Table 4 shows that the reduction in M&E services cost per percentage of glazed wall area was greatest for residential 3, again indicating that this project appeared to show different relationships to the others in the sample. All the residential projects comprised several individual blocks of accommodation. Values of the plan shape index, plan compactness parameter, square index, and perimeter index were calculated for each block, and an average value for the project was then calculated for each descriptor. When analysing the data for the residential projects it became apparent that the project averages for the above building form descriptors were misleading. All the above descriptors were based on the relationship between the area and perimeter of a building, but a ¯ oor area divided into, for example, four blocks, would require a different

Table 4 M&E services cost per percentage of glazed wall area, for residential buildings Project

Percentage of glazed wall area

Services cost per % Actual

glazed wall area Adjusted (factor 2)

Residential 1 Residential 2 Residential 3 Residential 4 Residential 5

10.18 10.37 12.60 15.42 19.42

£45 363 £44 019 £121 791 £56 615 £50 211

£43 844 £43 766 £85 938 £56 108 £49 160

489


Mechanical and electrical services amount of enclosing walls from the same area in a single building. Therefore it was decided to discount these building form descriptors from the analysis of the residential projects. Coincidentally the largest, most expensive and smallest, least expensive projects were both commercial buildings. Indeed it could be concluded that there was a vague relationship between M&E services cost and gross ¯ oor area for commercial buildings, as the project with the largest ¯ oor area also had the highest M&E services cost. The two industrial projects were not comparable in size nor in M&E services cost. The projects had gross ¯ oor areas of 1 540 m2 and 18 517 m2, and the actual M&E services cost tender ® gures were £425 897 and £1 798 059, respectively. The data indicated that percentage of glazed wall area was the building form descriptor with the strongest relationship with M&E services cost, for the industrial sample. When the M&E services cost was adjusted for project location, the maximum and minimum ® gures were £203 681 and £203 646, respectively, a difference of only £35 for each percentage of glazed area. In this sample the glazing ranged from 2.3% to 9.2% of external wall area. The relationships observed by inspection indicated that the actual M&E services cost per unit of the building form descriptors and ratios generally varied a great deal. The ranges were considered too wide for us to conclude with con® dence that particular relationships existed for each sample. Some of the adjustments applied for the factors affecting tender price levels did reduce the ranges for some variables. However, the variations in the relative successes of the different adjustments for each building form descriptor and ratio led to the conclusion that more accurate measures of the relationships were required to explain the apparent inconsistencies observed. It was considered useful to examine the average values of M&E services cost per unit of each variable, and the location of these values around the central tendency, for each of the samples. It was believed that this would allow relationships between the variables to be identi® ed with greater certainty than by inspection alone, and that the suitability of the various M&E services cost adjustments could be examined. Sample means and standard deviations The sample mean is the most common measure of location, and is the ordinary arithmetic average of a data set. One of the most important measures of variability was the sample standard deviation (Montgomery and Runger, 1994). The purpose of the adjustments applied to the M&E services cost tender ® gures was to facilitate

Table 5 Adjustments to M&E services cost that reduced the sample standard deviations for all variables Total sample

Residential sample

Commercial sample

Industrial sample

TPI Factor 2

TPI Number of storeys Type of building

TPI Number of storeys Type of building Factor 1 Factor 2

TPI Factor 2

comparisons between different types of project, tendered for under different market conditions. So far, the suitability of the various adjustments has not been discussed. It may be that some adjustments were useful only for certain building form descriptors or ratios, or for particular samples. For the purposes of this analysis, any adjustment that reduced the standard deviation of the actual M&E services cost data set was deemed to be useful as it reduced the variability in the relationships observed for the sample. In the sample containing all twelve projects, the sample standard deviation of the M&E services cost per m2 gross ¯ oor area was £80, with a sample mean of £192. The adjustments that resulted in smaller sample standard deviations were TPI (with S = 59), number of storeys (S = 79), and factor 2 (S = 63). This analysis was carried out for all building form descriptors and ratios, in all four samples. Table 5 shows the adjustments that reduced the standard deviations in each of the four samples. The residential sample had the smallest sample mean for actual M&E services cost per m2 gross ¯ oor area, which indicated that the residential projects were less highly serviced than the other building function samples. As the residential projects were not very highly serviced, they did not require a great deal of plant room space (the sample mean for percentage plant room ¯ oor area was 1.61% for residential projects, compared with 8.69% for commercial projects, 4.15% for industrial projects, and 4.98% of gross ¯ oor area for the total sample). The commercial sample had the lowest sample mean for actual M&E services cost per m2 plant room area, percentage plant room ¯ oor area, and m3 plant room volume. This was due to the relatively large spaces allocated to plant rooms, compared with the other samples. In the industrial sample, the standard deviations were generally large relative to the sample means, indicating a high degree of variability in the relationships observed in the sample. Exceptions, where sample standard deviations were small in relation to the sample means were


490 actual M&E services cost per m2 external wall area, actual M&E services cost per percentage glazed wall area, and actual M&E services cost per metre internal perimeter. These variables may prove useful for industrial buildings. Examination of the sample means and sample standard deviations for the building form descriptors and ratios, and the adjusted and actual M&E services cost values, revealed that there were differences in the relationships observed among the samples. This indicated that M&E services costs varied according to building function. The suitability of the various tender price factor adjustments was also examined. It was found that adjusting the actual M&E services cost tender ® gures for the tender price index was appropriate for all four samples; factor 2 was appropriate for all samples except residential; the number of storeys and type of building factor adjustments were suitable for use with residential and commercial projects, and factor 1 was suitable only for commercial buildings. Correlations Correlation is a dimensionless quantity that can be used to compare the linear relationships between pairs of variables in different units (Montgomery and Runger, 1994). Calculations of the strengths of the relationships between different sets of data, in this case tender costs of M&E services and various building form descriptors, were done mathematically with the correlation coef® cient. The correlation coef® cient r represents the strength of a relationship, ± 1 < r < + 1. If two sets of data have a perfect positive correlation, then as data set 1 is increased, data set 2 also would increase. Conversely, if a perfect negative correlation exists, then as data set 1 increased, data set 2 would decrease. If a correlation coef® cient were close to zero, there would be no apparent relationship between the data sets. Correlation analysis could identify only whether a linear relationship existed between two sets of data. There was no implication that a change in one variable caused a change in the other, both variables may have responded to a change in some other unobserved variable, or the observed relationship could be purely coincidental. Correlation analysis could not identify whether a nonlinear relationship existed between the variables, or whether the data fell into more than one pattern on a graph. The relationships identi® ed through correlation analysis were speci® c to the sample analysed. Correlational signi® cance indicated the percentage probability that the relationships observed were due to chance ¯ uctuations. The linear correlation coef® cients calculated were compared with crit-

Swaf® eld and Pasquire ical values from statistical tables, to identify whether the observed correlation was signi® cant. The total data set had a sample size of twelve, and therefore ten degrees of freedom. The 95% con® dence level was selected as adequate for the analysis. With ten degrees of freedom and p < 0.05, the critical value of the correlation coef® cient was 0.576. Therefore there was less than 5% chance of observing absolute values of correlation coef® cients greater than 0.576 by chance, and the relationship observed was considered signi® cant. For the total sample, correlation coef® cients were calculated for M&E services cost (the actual ® gure and the ® gures adjusted for the thirteen indices and factors described above and in Table 3) and twenty-two building form descriptors and ratios. A strong positive correlation was observed between gross ¯ oor area and M&E services cost, ranging from 0.9584 for the actual tender ® gure, to 0.8726 for the tender ® gure adjusted for the size of services contract factor. A signi® cant positive correlation for all M&E services cost adjusted data sets was observed also with usable area, circulation area, internal divisions, plant room area, plant room volume, usable height, total height, external wall area, glazed area, usable volume, and plant room area: usable area ratio. The strong positive correlations observed between M&E services cost and many of the building form descriptors indicated that as buildings increased in size (either in ¯ oor area, volume enclosed or height) the cost of the M&E services increased. This showed only that larger buildings had more expensive M&E services. It was decided that this relationship was not very useful in terms of identifying determinants of services cost, and that further analysis was required. Correlation coef® cients were calculated for M&E services cost per unit of each building form descriptor and ratio, such as per percentage of glazed wall area, or per metre of internal perimeter. This was done for the actual M&E services cost, and each of the adjusted tender ® gures, for the total sample, the residential sample, and the commercial sample. The industrial sample was too small for correlation analysis. Comparison of correlations observed for the three samples Tables 6 and 7 show summaries of the more signi® cant relationships observed from correlation analysis. Relationships between building form and function and M&E services cost appeared stronger for the commercial projects. However, none of the building form descriptors or ratios had perfect correlations with any of the M&E services cost per unit variable data sets, either actual or adjusted.

491

Mechanical and electrical services


Table 6 Most useful M&E services cost per unit variable data sets Total sample

Residential sample

M&E services cost per metre average storey height M&E services cost per wall:¯ oor ratio M&E services cost per m2 external wall area

M&E services cost usable height M&E services cost total height M&E services cost glazed wall area M&E services cost ¯ oor ratio

Commercial sample per metre

M&E services cost per wall:¯ oor ratio

per metre

M&E services cost per metre usable height

per percentage

M&E services cost per metre total height

per glazing:

M&E services cost glazed wall area M&E services cost M&E services cost wall area M&E services cost index M&E services cost

per percentage per glazing:¯ oor ratio per m2 external per plan compactness per square index

Table 7 Most useful building form descriptors and ratios Total sample

Residential sample

Commercial sample

Usable volume Usable area Internal cube Average storey height Horizontal distribution volume Wall:¯ oor ratio

Horizontal distribution volume Internal cube

External wall area Internal perimeter Usable volume Internal cube Plant room area Gross ¯ oor area Circulation area Horizontal distribution volume

Conclusions The research established that the M&E services cost (the research considered cost to the client, and used tender price as an approximation of cost) was indeed related to building function, because the relationships between M&E services cost and the building form descriptors and ratios analysed varied between the building function determined samples. It was observed that M&E services cost did not have precise linear relationships with building form and building function, and it was hypothesized that M&E services cost was not related solely to building form and function, and that the variations in cost relationships were due to the performance and quality of the M&E services. The study indicates that the analysis of M&E services cost in terms of building form descriptors is valid, but the commonly used building form descriptor gross ¯ oor area is not the most appropriate for M&E services cost estimates. Horizontal distribution volume and internal cube were the variables with the most signi® cant relationships with M&E services tender cost.

The sample sizes in the study were rather small, but the research methods described in this paper can be applied to larger samples in the future, to verify the preliminary ® ndings. There is scope also for development and testing of further descriptors, ratios and tender price adjustments. However, this work represents a contribution to knowledge in this important area.

References Banks, D.G. (1974) The uses of a length/breadth index and a plan shape index, Chartered Surveyor Building and Quantity Surveying Quarterly, 2(1), 15± 15. Brandon, P.S. (1978) A framework for cost exploration and strategic cost planning in design, Chartered Surveyor Building and Quantity Surveying Quarterly, 5(4), 60± 63. Brown, H.W. (1987) Predicting the elemental allocation of building costs by simulation with special reference to the costs of building services elements, Building Cost Modelling and Computers, Brandon, P.S. (ed.), E & FN Spon, London. BCIS (1969) Standard Form Of Cost Analysis, Principles, Instructions and De® nitions, Building Cost Information Service (reprinted 1995).


492 BCIS (1997a) Surveys of Tender Prices, Building Cost Information Service. BCIS (1997b) Indices and Forecasts, Building Cost Information Service. Chelmick, D.G. (1995) unpublished notes from interview carried out at ESL Engineering Services of® ce, London, 21 March. Crane, J. (1995) unpublished notes from interview carried out at Troup, Bywaters and Anders of® ce, London, 16 June. Ferry, D.J. and Brandon, P.S. (1991) Cost Planning Of Buildings, 6th Edn, BSP Professional Books, Oxford. Fletcher, L. (1968) Foreword, Cost Planning Mechanical and Electrical Engineering Services in Connection with Building Projects, Royal Institution of Chartered Surveyors, London. Kouskoulas, V. and Koehn, E. (1974) Predesign CostEstimation Function For Buildings, Journal of the Construction Division ASCE, 100(12), 589± 604. Leivers, M. (1995) Unpublished notes from interview carried out at Leivers Associates of® ce, Nottingham, 16 May.

Swaf® eld and Pasquire McLellan, A. (1995) Glaxo Wellcome Medicines Research Centre, Supplement to Architects’ Journal, October. Montgomery, D.C. and Runger, G.C. (1994) Applied Statistics and Probability For Engineers, Wiley, New York. RIBA (1969) Construction Indexing Manual, Royal Institute of British Architects, London. RICS (1982) Quantity Surveyors Practice, Pamphlet No. 2, Pre-contract Cost Control And Cost Planning, Quantity Surveyors Division, Royal Institution of Chartered Surveyors, London. Smith, M. (1995) Building Cost Survey ± Cost Model, Commercial Of® ces (Urban Environment), (The CIBSE Journal, September), published by Building Services Publications Ltd, London. Swaf® eld, L.M. and Pasquire, C.L. (1995) A critical analysis of building services cost prediction models, in ARCOM Eleventh Annual Conference, September 1995, pp. 424± 33. Swaf® eld, L.M. and Pasquire, C.L. (1996) A critique of mechanical and electrical services cost planning: existing methods and published information, Journal of Financial Management of Property and Construction, 1(3), 23± 41.