FHWA works to support statistical quality assurance techniques

While the feds are not dictating
“Percent Within Limits,” they want to
see a statistical technique of some kind
By Daniel C. Brown

The Federal Highway Administration has launched an effort to improve deficiencies in the materials quality assurance programs as practiced by the 50 states, the District of Columbia and Puerto Rico.

FHWA recently surveyed its division offices – one in each state – asking for information about that state’s materials acceptance methods, including quality control-quality assurance (QC-QA) for hot mix asphalt. “The data will help us direct our activities,” says Lee Gallivan, Asphalt Pavement Engineer in FHWA’s Office of Pavement Technology.

“Let’s say the survey identifies five deficiencies,” says Gallivan. “We’re going to use our resources to try to improve the deficiencies.” To help, FHWA may develop new workshops, training programs and conferences directed at bolstering states’ performances in material acceptance and quality assurance.

FHWA sees room for improvement in the area of consistent and proactive validations of contractors’ tests. Some states have stronger validation programs than others, and FHWA wants to help bolster the QA programs where needed.

FHWA not dictating
Some 30 to 35 states use contractors’ Quality Control tests for acceptance, and then do an independent validation of some kind. Another 25 to 30 states use contractor tests for acceptance and use a statistical technique called “Percent Within Limits (PWL)” to validate the consistency of the tests. And FHWA estimates that 20 to 25 states do all of their own testing and base their acceptance on those tests. (There is natural overlap among those three groups.)

In recent years, FHWA has encouraged the use of the PWL technique, and offers a workshop to explain it. More tests than may be conventional practice – of say, asphalt content, voids in the mix, in-place density and sometimes other properties – are required to figure the Percent Within Limits.

PWL is defined as the percentage of tests in a lot falling above the lower specification limit and beneath the upper specification limit. PWL uses the sample mean and sample standard deviation to estimate the percentage of the population (lot) that is within the specification limits. In theory, the use of the PWL method assumes that the population being sampled follows a normal bell curve, if test results are plotted on the X-axis and the percentage of tests are plotted on the Y-axis.

FHWA supports the use of some form of statistical method for acceptance of material and construction practices. “We are not dictating the use of Percent Within Limits,” says Gallivan. “There are other statistical techniques out there. We support statistical acceptance methods, of which PWL is one.”

States such as Maine, Kansas, Missouri and Arizona use the PWL technique – and they like the way it works. “The longer we’re involved in it, the more confident we are that it works,” says Joe Schroer, field materials engineer, Missouri Department of Transportation. “The contractors who used to take their lumps (in penalized payments) and move on have started making adjustments to their processes.”

Rick Barezinsky, field materials engineer with the Kansas Department of Transportation uses PWL. “I think the Percent Within Limits is an outstanding tool. It requires the contractor to have control over his operation,” says Barezinsky.

In Missouri, Schroer says the state uses PWL to calculate the pay factors for four properties: in-place density, asphalt content, voids in the mineral aggregate and lab air voids. There is a separate pay factor for tensile strength ratio, which indicates the asphalt’s susceptibility to stripping.

Missouri contractors typically run one each of the required quality control (QC) tests per sublot, or 500 to 1,000 tons of hot mix. The state runs its quality assurance (QA) tests at one per lot, or 4,000 tons. A typical specification for air voids is 4 percent, plus or minus 1 percent.

A contractor may, for example, find that his average QC air voids test is 3.7 percent, with a 0.2 percent standard deviation. The state allows plus or minus two standard deviations, which means the contractor’s population of air voids tests is said to run between 3.3 and 4.1 percent. If the state’s QA test falls between 3.3 and 4.1, says Schroer, “We say we fall into his population, and the material is accepted. But we have a caveat in our spec that says if the air voids fall below 2.5 percent, then the sublot has to be removed and replaced.”

If a contractor’s “Percent Within Limits,” based on his QC tests, hits 90 percent in Missouri, he gets full contract pay. Achieving 90 percent to 100 percent within limits will earn the contractor a bonus of up to 5 percent, on a graduated basis. By the same token, contractors are penalized on a graduated basis for achieving less than 90 Percent Within Limits. And anything below a total pay factor of 50 percent calls for removal and replacement.

A simpler system
Georgia does not use a PWL spec to control volumetrics, says Assistant State Materials Engineer Peter Wu, but the state does exact penalties for failure to meet specifications on gradations, binder content and density. Georgia allows contractors to vary from the target specification number by a “mean of deviations.”

For example, if the job mix formula calls for 6 percent binder, the amount of allowed deviation might be 0.72 percent. So if the contractor’s tests show a binder content of 5.3 percent, that’s the limit, and he will only receive 50 percent of pay. At 5.2 percent binder, the contractor must remove and replace. “The smaller the deviation from the job mix formula, the better chance he has to get full pay,” says Wu. “Our state law does not allow bonuses.

“I think Percent Within Limits is more strict,” says Wu. “They don’t just look at density and asphalt content, they look at VMA, VFA (Voids Filled with Asphalt), and the volumetrics. We give the contractor the opportunity to bring his mix back within compliance. If you see your binder content going down, from 5.4 to 5.3 to 5.2, you need to bring the mix back. We stop the plant and work with the contractor. But with PWL by the time they test it the mix is on the road. It’s too late.”

Is it broken?
Iowa is another state not currently using a PWL spec. In Iowa, the Quality Control/Quality Assurance (QC/QA) program allows the contractor’s test results to be used for pay, while split samples go to the state’s district office for verification.

“We are moving in the direction of Percent Within Limits,” says Mike Kvach, Executive Vice President of the Asphalt Paving Association of Iowa. Recently a memo from the Iowa Department of Transportation to its district offices asked the districts to meet with contractors and identify “shadow projects” for an effort called “Quality Mix Asphalt II.” The memo asks for contractors and districts to volunteer to provide additional information on some projects about the impact of proposed changes to Iowa’s hot mix asphalt acceptance specifications.

“The current (specifications) will be used for project acceptance, but for the shadow projects additional information will be gathered,” says Kvach. “QMA II is the next step toward Percent Within Limits.

“Industry is not convinced that what we have is broken,” says Kvach. “However, we sense that there is pressure on the state from the federal level to adopt a more ‘statistically’ based QC/QA program. We have been working as partners with our state DOT to assure that the end result is a specification that delivers quality and consistency – yet remains practical to achieve in the field.

“The danger is that we could end up with a spec that is not attainable or doesn’t reward the contractor for improved quality,” says Kvach. “If we rush into this, we could end up losing some of the practicality of our hot mix paving process.”

Thorough testing process
In Maine, the Department of Transportation applies the PWL technique to seven HMA properties: aggregate gradation, binder content, lab voids, Voids in the Mineral Aggregate, Voids Filled with Asphalt, the ratio of fines to effective binder and in-place density.

“We monitor the quality level on all of those properties, and we can initiate a shut-down if any one of them drops below a certain level,” says Richard Bradbury, quality assurance engineer for the Maine DOT. Pay adjustments, however, are made for four properties: binder content, air voids, VMA and in-place density. Maine uses the AASHTO-suggested formula for pay factor, which is: Pay Factor = (PWL X 0.5) + 55. So if your material is 100 Percent Within Limits, your pay factor is 50 + 55, or 105 percent.

“Percent Within Limits works very well for us,” says Bradbury. “It allows us to set realistic specification limits. In the past our specifications were based on historical values. I am responsible for looking at our spec limits every three years and I make sure that our spec limits are fair and reasonable, based on what the industry is providing.

“PWL is a more fair and equitable way of doing things,” says Bradbury. “Before we had PWL, we had an inspector in the plant helping the contractor to make adjustments. It wasn’t as scientific a way to do things. A contractor should control his own quality.

“By using a statistical control method, we get a better measure of the actual quality of the asphalt mixtures,” says Bradbury. “Before, we didn’t have a way to measure how much of the material was acceptable and how much was not acceptable. Now we have a way to measure the quantity that is acceptable. And we can compare the quality we receive from contractors across the state.”

What about timing of the tests? Bradbury says the state can turn around results on a project by the morning of the second day after the tests are run. Contractors are required to run the same tests the state does. For densities, contractors use nuclear density gauges, and the state cuts cores.

All tests by Arizona
In Arizona, either the state or a lab paid by the Department of Transportation runs all acceptance tests. Pay factors are based on gradation (four sieve sizes); asphalt content, lab air voids, pavement thickness, and compaction. The state cuts cores to determine density.

As do many states, Arizona weights its pay factors to place more value on the hot mix properties it considers more important, says Julie Kliewer, P.E., PhD, pavement materials testing engineer. In Arizona – that’s lab air voids. Pay is calculated by combining the pay factors for gradations and binder content, then adding that bonus or penalty to the lab air voids pay thus giving it equal weight to the other mix properties. Penalties or incentives are figured as set amounts per ton, not percentages of the contract.

Kansas runs a list of 12 tests on aggregates and mixtures, including VMA, VFA, coarse aggregate angularity, fine aggregate angularity, sand equivalency, a tensile strength ratio test and more. If one of them falls out of spec, the state can shut down the contractor.

“We verify those using an F&t analysis, which is a statistical tool,” says Barezinsky of Kansas, who teaches a two-day course on statistical material control. “For air voids, we use a Percent Within Limits analysis method. And for thick overlays, we use a PWL for densities.”

FHWA’s Gallivan estimates three of the nation’s top-ranking statistical control programs are run by Maine, Kansas and Washington state. Maine and Kansas both use the PWL technique for acceptance. “The PWL spec requires contractors to run at a fairly tight deviation to get our incentive payment,” says Kansas’ Barezinsky. “If they don’t run consistently, they run the risk of a payment penalty. Most of the time they do fairly well, but every contractor has a bad day or two.”

Dan Brown is the principal of TechniComm.