Improving the Accuracy of Early Cost Estimates for Federal Construction Projects (1990)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

APPENDIX D 33 Appendix D Current Procedures for Preparing Early Estimates Prepared by Michael Morris* Many different methods and techniques are used to prepare early estimates for construction; that is the pre- programming, program, concept/schematic, and design development estimates. This appendix discribes the most commonly used methods. Some of the methods discussed are also used to prepare detailed owner (government) estimates and estimates to compare with contractor bids; such as, working drawing and contact document estimates. However, since the focus of the appendix is on early estimates, the methods are discussed only in that context. Techniques for preparing early estimates can be categorized in various ways. In this appendix, the techniques are grouped and discussed under four broad headings: single unit estimates, multielement estimates, parametric estimates, and range estimates. Information for the discussion was obtained from many sources, but especially Adrian (1982) and Bower (1984). SINGLE UNIT ESTIMATES In single unit estimates, the cost of a facility is calculated on the basis of a unit of measurement, which may be expressed in terms of the functional use of the facility, areas and volume, or factoring. Estimates Based on Function-of-Use Units Among the function-of-use units sometimes used to prepare early estimates are the number of bedrooms in a hotel, the number of seats in a theatre, the number of beds in a hospital, and the number of parking spaces in a parking garage. Function-of-use estimates are prepared by multiplying the number of units to be included in a proposed facility by an average construction cost per unit (e.g., dollars per bed). Such estimates can be prepared very easily and quickly if an estimator has the appropriate historical data— and when an owner constructs a particular type of facility on a regular basis, historical cost data related to a function-of-use unit for that facility can be accumulated easily. However, even with ample historical data, estimates based on functionof-use units frequently are unreliable since costs per unit are subject to wide variation. It is generally agreed that estimates based on function-of-use units ought to be used only for very preliminary planning purposes (i.e., pre-programming estimates) and only by people who are thoroughly familiar with the type of facility in question and recognize the limitations of the data. Estimates Based on Areas or Volume Generally, one of the following three units are used to prepare area or volume budget estimates: floor area, building surface area, or building volume. Floor area is the most widely used unit for *Michael Morris is President of Hanscomb Associates, Inc. 

APPENDIX D 34 preparing early estimates in most parts of the world. Estimates based on floor area generally are prepared by measuring the gross floor area of the facility and multiplying the area by an average unit cost (e.g., dollars per gross square foot). There is an abundance of data available on construction costs per square foot for many types of facility, and government agencies, developers, contractors, and cost consultants collect, store, and use such cost data routinely. In preparing areabased estimates, the estimator must be concerned with the accuracy of available unit cost data and have the necessary estimating skills to select the most appropriate rate for the building being estimated. Judgment is required because most cost data at this level includes little information on the characteristics of the buildings included in the historical data base. Area-based cost data is published by several commercial houses and is widely used by owners and estimators who do not have data bases of their own. These data, which are typically generated annually, give median costs per square foot for various building types. The unit cost for each building type is related to a “typical” building of a particular size. If the size of the building being estimated is different from the size of the typical building, the unit costs may be adjusted using a nomograph called a “square foot project size modifier.” To improve the accuracy of floor-area based estimates, estimators sometimes use different unit costs per square foot for different functional areas of a building. This is known as the functional area method of estimating and the purpose is to account for the fact that in some types of buildings (such as hospitals) the cost per square foot to construct various departments may be significantly different. However, obtaining accurate cost data broken down by functional area can be difficult, and developing estimates using such data requires considerable skill. One problem with all floor-area based estimates is that there are no universally accepted rules for measuring building areas. The American Institute of Architects has developed measuring rules, but many organizations— including most federal agencies—do not follow them. In fact federal agencies cannot agree among themselves on definitions of building areas; in a recent Federal Construction Council study (Consulting Committee on Planning and Design Terminology, 1988) it was found that various agencies use at least 13 different terms to define the areas of their buildings. Clearly, in the absence of standards on measuring and defining areas, area-based cost data must be used with caution. The building surface area method of estimating is similar to the floor area method. However, whereas with the floor area method the estimate is based on the area of just the horizontal floor surface of the building, with the building surface method both vertical and horizontal surfaces are considered. To prepare an estimate based on a building surface areas, the surface areas of all floors, roofs, and external and internal walls are measured (but only one surface of an element is measured). An average unit cost rate for all surfaces is then selected and multiplied by the measured surface area to compute the total cost of the facility. There are also variations on this theme; for example, in some cases the areas of the various elements are multiplied by weighing factors to reflect the differences in their construction costs. The building surface area method is not widely used because reliable cost data of the type needed is often not available. Furthermore, if drawings of the building are sufficiently well-developed to allow use of this method of estimating, most estimators tend to use a multi-element approach (see below) for which unit cost data are more readily available. The building volume method of estimating also is similar to the floor area method. The main difference is that the height as well as the floor area of the building is considered. The building volume method is not nearly as popular as the floor area method, but is sometimes used where buildings of the same type can have significantly different floor to ceiling heights (e.g., hangars, hospitals, and warehouses). The building volume method—like the building surface area method—is used infrequently partly because of a lack of cost data; however, some published cost-per-cubic-foot data is available in the market place. The Factoring Method of Preparing Budget Estimates The factoring method of preparing early estimates is most often used for major manufacturing facilities where the cost of a single component, for example equipment, is the most significant cost item. Costs of the various components of the building required to house and support the equipment are assumed to be fixed percentages (factors) of the 

APPENDIX D 35 cost of that equipment, as illustrated in the following example. Cost of Equipment: $2,000,000 Cost of construction and support systems expressed as factors of the equipment costs: Element/Component Factor Estimated Costs Architectural/Structural .30 $600,000 Mechanical .25 450,000 Electrical .10 200,000 Equipment Installations .15 300,000 General Conditions .10 200,000 Total Estimated Cost $1,750,000 The type of cost data used with the factoring method is relatively inexpensive to collect and use, and the factoring method can produce reasonably accurate estimates provided a good data base is available. The factoring method does not, however, lend itself to the type of facilities ordinarily constructed by the federal agencies. MULTI-ELEMENT ESTIMATES The multi-element approach is a popular method of preparing cost estimates for construction: First, each of the various elements, systems, and components of the proposed facility are identified, sized, and priced separately in accordance with recognized procedures; next, the costs of the individual items are added to determine the total direct cost of the facility; finally, allowances are added for overhead, profit, and contingencies to determine the overall estimated cost of the project. Most estimates of the multi-element type are based on one of two recognized formats: an elemental format in which most costs are related to the systems and physical elements that make up a building; and a trade format, in which costs are broken down by specification sections, most of which are related to construction trades or materials. Elemental Format Probably the most widely used elemental format is Uniformat (a contraction of “uniform” and “format”), originally developed by the American Institute of Architects (AIA). It was later modified and adopted by the General Services Administration (GSA). The GSA version is now a nationally recognized method for analyzing building construction costs on the basis of 12 standard building systems or elements: foundation, substructure, super-structure, exterior closure, roofing, interior construction, conveying systems, mechanical, electrical, general conditions and profit, equipment, and site work. As developed by GSA, each of the 12 Uniformat elements can be further broken down into sub-elements, and each sub-element can be further subdivided into components to permit the preparation of highly detailed estimates. However, in practice, most estimators use the Uniformat breakdown for preparing early estimates and the CSI approach (see below) for preparing working drawing estimates. Some commercial houses and several federal agencies publish generalized cost data for early estimating purposes on the basis of the 12 Uniformat elements, but also have available detailed cost data for preparing more detailed estimates in accordance with the CSI format. The Uniformat approach has been widely adopted for the preparation of early estimates because owners, architects, engineers, and others involved in making broad decisions about construction tend to relate more to the cost elements used in Uniformat than to the work items in the CSI format. The usefulness of the Uniformat approach has been enhanced by the development of computer programs that permit cost data in the CSI format to be sorted into Uniformat elements and vice versa. CSI Format The most widely used format for storing and presenting construction cost data is the 16-division specification format of the Construction Specifications Institute: general requirements, site works, concrete, masonry, metals, wood and plastics, thermal and moisture protection, doors and windows, finishes, specialties, equipment, furnishings, special construction, conveying systems, mechanical, and electrical.* By design, the CSI format reflects the scheme used by most contractors to organize and manage construction projects, account for costs, and award subcontracts. Consequently, the CSI format is widely used by contractors to prepare bids and by *The 16-division specification format of CSI is based on the 16-division “Uniform Construction Index” that was jointly developed in the late 1960s by a number of organizations including the American Institute of Architects, the Associated General Contractors, and CSI. 

APPENDIX D 36 estimators to store cost data for working drawing estimates. However, while there is considerable historical cost data available in the CSI format on various building types, it is often of little value for early estimating purposes because it is usually in such detail that it cannot be used until the design of the facility has been developed to a significant degree. A detailed estimate using the CSI format is prepared by measurement of the quantities for all labor, materials, and equipment required for each item included in each of the 16 divisions. These quantities are then priced at appropriate rates, extended, and totaled. Allowances are then included for general conditions, overheads, profit, and contingencies to arrive at a total estimated cost. The preparation of such estimates, generally known as the quantity survey method, is labor-intensive and needs time to complete. It should be noted that detailed working drawing estimates can be prepared in CSI or elemental format or indeed in other formats to suit an owner's code of accounts. As mentioned previously, the managers and professionals who make decisions in the early stages of a project do not think in terms of the work items in the CSI format. Rather, they tend to think in terms of systems, like those in the Uniformat. Thus, whereas detailed estimates are usually in the CSI Format, program and concept/schematic estimates tend to be in Uniformat. PARAMETRIC ESTIMATES To a certain extent all methods of estimating are parametric in that they are based on the use of cost parameters. However, the term “parametric estimating” is generally understood to mean the technique of developing estimates based on a limited number of important features that are the major cost drivers of an estimate. Parametric estimating is most applicable to relatively standard facilities since the starting point is a data base containing detailed estimates of various specific facilities. The premise underlying parametric estimating is that the cost of the facilities in the data base will vary as a function of certain values (parameters). Thus, by assigning new values to the parameters associated with the detailed estimate from a particular facility, a new detailed estimate for that facility can be generated. The concept is sound and produces reasonably accurate estimates provided the algorithms and statistical data used in connection with the parameters are accurate and extrapolations of the estimates in the data base are not excessive. The major advantage of parametric estimating is that it provides detailed cost breakdowns—in either Uniformat or CSI format depending on how the prototype estimates in the data base are formatted—quickly and at relatively low cost with only limited analysis of the facility to be constructed. Thus, with parametric estimating, budget estimates can be prepared that include similar detail as working drawing estimates. The disadvantages of parametric estimating are that it can be used only for facilities similar to the facilities for which there are estimates in the data base, and the computations are performed by computer, making it very difficult for estimators to verify the results. In addition, many parametric estimating systems are proprietary, and developers will not divulge the algorithms used; consequently, their validity must be taken on faith. The U.S. Air Force has developed a parametric estimating system called the Construction Cost Management Analysis System (CCMAS) for estimating costs of various types of Air Force facilities that are constructed on a regular basis. The developers of the system, the Construction Cost Management group at Tyndall Air Force Base, report that the system has been tested and found accurate and reliable (Bridges and Gregory, 1987). PROBABILISTIC ESTIMATING AND RANGE ESTIMATING Construction cost estimating traditionally has been treated as a determinate problem; that is, a problem in which the answer can be expressed as a specific, definite value. However, since a cost estimate is really a prediction of what an item or group of items will cost in the future, many assumptions must be made in preparing an estimate, which introduces a degree of uncertainty into the process. In the case of early estimates, which usually are prepared before most design decisions have been made (and often many months before bids are received), the level of uncertainty may be high. Traditionally, estimators and owners have dealt with uncertainty about the accuracy of estimates through the use of contingency factors, which in essence provide funds to cover cost overruns up to a certain amount. Although the contingency-factor approach has worked reasonably well, in recent years a number of estimators have developed alternative methods of quantifying the uncertainty that is inherent in almost all cost estimates. These methods are usually referred to as either range estimat 

APPENDIX D 37 ing (Curran, 1988) or probabilistic estimating (Consulting Committee on Cost Engineering, 1983). While there are significant differences in the various range estimating and probabilistic estimating techniques that have been developed, they tend to have several features in common; specifically: (1) they require that an estimate be made of the potential variability of each element in an estimate; (2) they employ the laws of probability to determine the impact of possible variations in the cost of individual elements on overall costs; (3) they require the use of a computer; and (4) they present the results in the form of a histogram or a cumulative distribution showing either the probability of various estimates proving to be the actual cost of the project, or the probability of cost overruns of various magnitudes. Opinions vary on the value of range estimates for budgeting purposes. A number of users have enthusiastically endorsed the range estimating concept (see Curran 1988). However, several federal agencies that used range estimating on a trial basis encountered opposition from estimators on the grounds that the range estimating process as too time-consuming and from managers on the grounds that they did not want more complexity in the decision-making process (see consulting Committee on Cost Engineering, 1983). 

APPENDIX D 38