Highway Quality Compendium
Dispelling Highway Construction Myths
by Peter Kopac
Examining a few commonly believed half-truths may help materials, structural, and pavements engineers develop sound and effective quality assurance programs.
Sometimes when people hear or read an idea often enough, it becomes accepted as fact and ingrained as a self-evident truth. Invariably, these notions are passed on to others, and soon no one questions them any more. "Man was not meant to fly" was accepted as fact for centuries. But because a few people did not accept that belief, they developed an important means of transportation.
Part art and part science, the discipline of quality assurance for highway construction abounds with half-truths, myths, and misconceptions. These myths typically originate from well-meaning sources. Some myths serve a worthy function by simplifying the difficult to make it more understandable. However, on the negative side, myths:
- May be partly true, but not all of the time, so they can lead to an incomplete understanding of important concepts
- Encourage decisionmaking as a seat-of-the-pants approach rather than one based on facts and data
- Leave a narrow, rigid impression that stifles creativity
For quality assurance to keep progressing, highway agencies and contractors might examine several key myths that have become firmly entrenched as truth. Dispelling the more persistent myths that have gained a strong foothold can help engineers develop a better understanding of the concepts and principles needed to produce and implement sound and effective quality assurance specifications and programs.
Myth No. 1-You Can't Inspect Quality Into the Product
This assertion is often made to reinforce the viewpoint that only the highway contractor can build quality into the product. This myth contends that it is too late for the agency to improve quality once the contractor's product has been submitted for acceptance.
This is true under only one scenario, that in which the contractor's product must be accepted as is, with a pay decrease being the agency's sole recourse for deficient quality. Typically, that is not the case in highway construction.
During construction, it is a daily occurrence for agency inspectors to reject truckloads of portland cement or asphalt concrete. Although it does not happen too often, the contractor is required to remove and replace entire lots exhibiting very low quality. In some cases, inspections can lead to inplace correcting and reworking of unacceptable product. Some examples are pavement surface grinding to improve smoothness, scarification (loosening) and addition of material to increase base course thickness, and removal and replacement of segregated or honeycombed areas to improve durability and appearance.
One way to measure the effectiveness of a construction acceptance plan is by how much quality has improved as a result of inspection. This measure of the effectiveness of an acceptance plan is based on the assumption that a person can inspect quality into the product. Practitioners of highway construction quality assurance generally have overlooked this measure. It calls for determining the difference or ratio between the average incoming quality submitted by the contractor and the average outgoing quality after inspection and rectification.
Some acceptance plans are designed to control or optimize the average outgoing quality. These plans depend on an initial sampling inspection to estimate the number of defectives in a lot. When the number is high, 100-percent inspection is conducted. Any accepted lot is improved by the elimination or correction of defectives found during the inspection.
Such acceptance plans have been applied mostly in industrial situations, but they have the potential for various applications during highway construction. To control the average outgoing quality, either the highway agency or the contractor can use the acceptance plan to inspect such items as masonry units, reinforcement bars, dowel bars, guardrail, right-of-way fence-indeed any product delivered in bulk.
Myth No. 2-The Ultimate Goal Is Product Performance
Agencies often present this performance argument in support of warranty specifications that hold the contractor responsible for a specified level of pavement performance, as opposed to a specified level of materials and construction quality. The argument maintains that although quality and performance go together and are both important, performance should be specified because it is the agency's ultimate goal.
But is it really? Striving for high pavement performance without sufficient regard for the cost to achieve that performance level is not in keeping with the best interests of the road user. Ideally, an agency's goal should be neither maximizing quality nor maximizing performance, but instead should consider minimizing life cycle costs (including user costs due to crashes, delays, noise, and so forth).
Although most developers of warranty specifications consider costs in establishing performance thresholds, the established thresholds may not be at the optimum values to also minimize life cycle costs. This issue can be addressed by developing warranty specifications that seek to minimize life cycle costs. Another approach would be for an agency to directly specify the life cycle costs desired.
To assist agencies, the Federal Highway Administration (FHWA) developed guidance for performance-related specifications that focus on minimizing life cycle costs. The guide specifications are based on pavement performance models that convert various levels of construction quality into estimated life cycle costs. The standard way for an agency to apply these specifications is to specify the desired construction quality level and hence the desired postconstruction life cycle cost of the pavement. The paving contractor can obtain a pay increase by providing a quality level that results in a lower-than-specified postconstruction estimate of life cycle cost.
Another more innovative way to apply these specifications might be for each potential paving contractor to submit a target life cycle cost as a bid. The contractor with the lowest bid is awarded the project. Here again, the contractor can obtain a pay increase by achieving a lower-than-targeted estimate of life cycle cost.
Much knowledge has been gained in recent years on the relationship between construction quality and construction performance. Today, the weak link in the advancement of quality assurance seems to be lack of knowledge about the costs of quality and performance. Any quality assurance effort to specify or increase the level of construction performance needs to pay proper attention to the cost elements.
Myth No. 3-With the Evolution of Specifications Come Continual Improvements in Quality
Specifications evolve as a result of two general factors:
- Advances in engineers' knowledge of and experience with the product being specified, including the materials, processes, and equipment
- Political, economic, and societal demands placed on the product or on those responsible for the product
The first factor leads to product quality improvements and/or to more cost effective ways to produce the product. The second factor, however, does not necessarily lead to quality improvements.
There are many examples of specifications evolving due to items that fit into the second factor. One example is the increased use of waste or recycled materials in highway construction. Waste and recycling programs may be driven more by environmental concerns than by the need for quality and performance improvements. The dwindling supply of natural resources has given little choice but to view entire pavements as potential aggregate sources that can be recycled.
The seemingly continuous downsizing of agency personnel is another example where specifications are forced to evolve with little regard for their impact on quality. Faced with personnel reductions, some agencies are attempting to "do more with less." Other agencies are delegating acceptance testing duties to the contractor in charge of construction and in effect "doing less with less." Such agencies have placed themselves in the uncomfortable position where their role is primarily to validate the contractors' test results.
Yet another example is the steady increase of new dryer drum hot-mix asphalt plants since the 1970s, resulting in a decline in the percentage of batch plants. Dryer drum plants offer important advantages over batch plants in producing more tonnage per hour, operating more economically, and emitting fewer pollutants. However, dryer drum plants cannot supply several different mix designs in the same production run easily (that is, without multiple additional silos, which would not be feasible for portability). Although they cost more to operate, batch plants also have advantages: They have the ability to switch mix specifications-in the middle of a truck if needed-and there is more control over individual components of the hot mix, such as aggregate size and gradation control. In the 1997 Roads & Bridges article, "How to Choose the Right Asphalt Plant," the author writes, "A batch plant's strength lies in its ability to make salable hot mix out of almost any reasonable stockpile of aggregate."
The above examples illustrate how political, economic, and societal demands often conflict with and outweigh quality considerations. As specifications evolve partly in response to these demands, the quality being specified does not necessarily increase continuously.
Myth No. 4-Quality Assurance Specifications Provide Higher Quality Than Method Specifications
This too is not necessarily true. The truth is that quality assurance specifications have the potential to yield a higher quality.
Under method specifications, the contractor follows agency-prescribed methods while using agency-approved materials and equipment. The resulting construction quality is thus dependent on the methods, materials, and equipment described in the specifications. The resulting quality is the minimum quality level described in those method specifications. Under these specifications, the low-bid contractor has no incentive to use better methods or materials that will result in a higher quality than that corresponding to the specified methods and materials.
On the other hand, the contractor working under quality assurance specifications typically does have an incentive, in the form of positive/negative pay adjustment provisions, to provide as high a quality as is profitable. Thus, assuming the same minimum acceptable quality level is specified, properly developed quality assurance specifications can result in higher quality than method specifications.
The problem is that the vast majority, if not all, of comparisons of the quality produced under different types of specifications (whether they be method, quality assurance, or warranty) are made without ascertaining the quality levels being specified. Apples are not being compared to apples. It stands to reason that if the quality level specified under method specifications is greater than that specified under quality assurance specifications, the method specifications will result in a higher quality, unless the difference in specified levels is so small that the incentive effect becomes significant. For example, if an agency specifies methods under the method specification that lead to an initial International Roughness Index (IRI) of 40, that method specification is likely to result in better smoothness (lower IRI) than a quality assurance specification designed to provide an initial IRI of 80.
The whole point of specifications is that they are supposed to tell the contractor what the agency wants. If the agency wants higher quality (within reason), it should be able to describe that desired quality level regardless of the type of specifications it chooses to employ. Certainly, the best indicator of the quality to be achieved on a project is the quality level being specified, not the type of specifications. The easiest and most straightforward way for an agency to obtain high quality construction is simply to ask for it-that is, to specify it.
Myth No. 5-Percent Within Limits (PWL) Is the Ideal Quality Measure
The PWL, or its complement the Percent Defective (PD), is currently the recommended statistical measure of specified materials and construction quality. Ninety percent within limits (that is, 10 percent defective) generally is considered an acceptable quality level. The basis for recommending PWL is that it nicely combines two important parameters, the mean and the standard deviation, into a single quality measure. For most acceptance quality characteristics, PWL provides a better measure of specified quality than the other single measures, such as the average, moving average, average absolute deviation, conformal index, and various other quality indexes.
However, PWL is far from an ideal measure. Its major shortcoming is that a given specified PWL can describe an infinite number of populations. In the Possible Asphalt Content Populations chart, all three represented populations have 15 percent defective, yet each will result in a different pavement performance because too much asphalt content leads to bleeding and loss of skid resistance, and too little asphalt content leads to early deterioration.
This leads one to conclude that PWL does not correlate strongly with performance. Thus, the use of the PWL quality measure is problematic for highway agencies seeking to develop pay adjustment schedules that truly relate measured quality to expected performance and to the contractor's pay factor.
Agencies need to clearly specify the quality level they want. Instead of trying to specify through an ambiguous single quality measure such as PWL, agencies should consider another option-two quality measures. The typical way to describe a normal distribution population is by its mean and its standard deviation. Several agencies have used this approach successfully in developing pay adjustment schedules that relate various combinations of mean and standard deviation to contractor pay factors. FHWA's Guide to Developing Performance-Related Specifications for PCC Pavements (FHWA-RD-98-155) describes such an approach.
Myth No. 6-All Pay Adjustment Schedules Provide an Incentive for Quality
The Transportation Research Board's Circular Number E-C037, "Glossary of Highway Quality Assurance Terms," makes a distinction between a pay adjustment schedule and an incentive/disincentive provision. According to the glossary, an incentive/disincentive provision is "a pay adjustment schedule that functions to motivate the contractor to provide a high level of quality." The glossary further states: "A pay adjustment schedule, even one that provides for pay increases, is not necessarily an incentive/disincentive provision, as all possible pay adjustments may not be of sufficient magnitude to motivate the contractor toward high quality."
A pay adjustment schedule must be developed properly if it is intended to serve as an incentive/disincentive provision. To develop the provision for a given quality characteristic such as strength, asphalt content, or smoothness, the agency should first understand the relationship between quality and the cost to the contractor of achieving that quality. Unless the agency has a feel for how much it costs contractors to achieve higher quality strength, for example, the agency does not know whether its pay schedule for strength provides the contractor sufficient incentive to produce a higher quality strength.
What complicates the matter, however, is that most agencies develop not one but several schedules designed to function together as a pay adjustment system. These agencies typically employ various composite equations to combine the calculated pay factors for each individual quality characteristic. A composite pay factor equation often includes caps to define the minimum and/or maximum composite pay factor allowed. In some cases, the inclusion of caps makes it more profitable for a contractor to target a decreased quality level for one or more individual quality characteristics and still be assured of obtaining a high composite pay factor.
Agencies that have pay adjustment schedules should do a thorough check to determine whether they are true incentive/disincentive provisions. The University of Florida currently is developing computer software for that purpose, called Prob.O.Prof, an acronym for Probabilistic Optimization for Profit. Using the software, a State or local department of transportation (DOT) should be able to identify those pay adjustment schedules and systems that may not provide an incentive for quality. Contractors also should be able to use the software to establish optimum target values that will result in maximum profit.
Myth No. 7-Agencies Need Not Measure Materials and Construction Quality on a Warranty Project
This is totally untrue. The use of warranty specifications does not eliminate the need for agencies to measure quality during construction. In fact, during this time period when new warranty specifications are being developed and tried, the need for quality testing is of utmost importance.
Many agencies currently using warranty specifications already require some minimal testing to establish whether the contractor has achieved the specified quality levels on key characteristics, such as thickness or strength, and therefore can be relieved of the responsibility for certain corrective or remedial actions that may be necessary later. Other important reasons why agencies should measure quality on warranty projects are the following:
- To determine whether quality is improving. Are materials and construction quality better, worse, or the same under new warranty specifications or under the current specifications? Do warranty specifications, once the contractor has sufficient experience with them, promote continuous quality improvement?
- To establish or validate the relationship between quality and performance. What levels of quality need to be built into the construction to achieve various levels of product performance?
- To obtain an estimate of how the delivered product will perform. When can maintenance and repairs be expected? How long will the product last?
- To identify the possible causes of poor performance. What went wrong? What test data will the dispute resolution team need?
The introduction of warranty specifications has created a shift from agency acceptance of product quality during construction to agency acceptance of product condition at the end of the warranty period. Under warranty specifications, the agency needs to establish a strong independent quality assurance program, or, better yet, a strong independent performance assurance program. Either way, agency testing of quality during construction is something that agencies cannot afford to do without.
Myth No. 8-Buyer's and Seller's Risks Should Be Balanced
Buyer's (agency) and seller's (contractor) risks are nothing more than probabilities. Simply stated, the buyer's risk is the probability that poor-quality construction will be accepted, and the seller's risk is the probability that good-quality construction will be rejected. For the development of a construction acceptance plan, the objective is not to balance the risks themselves but to balance the multiplied product of the risks and their consequences.
Little guidance is available on the relative appropriate size of buyer's and seller's risks. The National Cooperative Highway Research Program (NCHRP) Report 17, Development of Guidelines for Practical and Realistic Construction Specifications, suggests the risks shown in a Buyer's and Seller's Risks table. The table has served to generally define what can be considered appropriate levels of risk, but one should note:
- The buyer's and seller's risks are not balanced (that is, not equal).
- The balancing involves the size of the risks and their consequences. Highway agencies should be able to better pinpoint the optimum risks for specific quality characteristics and specific consequences. To do so requires an understanding of operating characteristic (OC) curves.
A simple subjective way to arrive at the "optimum" risks is for an agency and its contractors to examine together the OC curves for a given acceptance plan. If either party is not satisfied with the risks, adjustments can be made to the plan, resulting in revised OC curves. An iterative process of adjusting the plan and examining its OC curves can be followed until a satisfactory compromise has been reached.
To arrive at the optimum risks more objectively should not be too difficult. The process involves economic decision theory and, like the subjective approach, requires an understanding of OC curves (probabilities) and economics (cost of consequences). Through such an approach, an agency also can determine the optimum sample size (that is, the number of samples, n) necessary to minimize the cost of its acceptance plan. For a given acceptance plan, the larger the n, the greater the costs associated with testing; however, the larger the n, the lower the buyer's and seller's risks (that is, the lower the probability of incurring costs associated with undesirable consequences).
The field of quality assurance is full of myths. A few of the more common ones, discussed here, are all simple or absolute statements that sound good or look good at first glance. Rather than fall into the trap of believing all simple and catchy statements, the quality assurance practitioner should give such statements due consideration and thought. Once a myth is identified, a better conceptual understanding can be reached, promoting more of the out-of-the-box thinking that is necessary for quality assurance to keep progressing.
1. Hoerner, T.E., and Darter, M.I. Guide to Developing Performance-Related Specifications for PCC Pavements, Volume I: Final Report. Report No. FHWA-RD-98-155. Washington, DC: Federal Highway Administration, February 1999.
2. Burati, J.L., et al. Optimal Procedures for Quality Assurance Specifications. Report No. FHWA-RD-02-095. Washington, DC: Federal Highway Administration, April 2003.
3. Indiana Department of Transportation. Performance-Related Specifications for Use in Determining Pay Factor Adjustments. Prepared for Project R-25175, Interstate 65. Clarksville, IN, March 2002.
4. Committee on Management of Quality Assurance. Transportation Research Circular Number E-C037, Glossary of Highway Quality Assurance Terms. Washington, DC: Transportation Research Board, April 2002.
5. Vidalis, S. Investigation of the Relationship Between the Quality and Cost of PCC Pavement Construction. A Ph.D. dissertation for the University of Florida. Expected to be completed in May 2005.
6. Miller-Warden Associates. National Cooperative Highway Research Program Report 17, Development of Guidelines for Practical and Realistic Construction Specifications. Washington, DC: National Research Council, 1965.
Peter A. Kopac is a research highway engineer on the Pavement Materials and Construction Team of FHWA's Office of Infrastructure Research and Development. He has more than 35 years of highway-related experience, including 28 years with FHWA. Kopac's primary research focus has been on quality assurance systems. He has assisted numerous agencies in developing, reviewing, and analyzing their quality assurance specifications. He is also an active member of Transportation Research Board Committee AFH20, Management of Quality Assurance.
Reprinted from Public Roads, May/June 2005.