Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 212


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 211
Field Adjustments and Quality Assurance of HMA Mixtures 211 be increased by increasing the proportion of coarse aggregate. For mixtures near the maximum density gradation, air void content can be increased by increasing the proportion of fine or coarse aggregate. The rules given above are general and, in practice, there are many exceptions; plant operators should use their experience in adjusting aggregate proportions during field production to obtain the target air void content and VMA. The general mix design procedure described in Chapter 8 of this manual should also provide information on how aggregate gradation changes will affect air void content and VMA, which can be used as a guide when making adjustments during field production. It should be emphasized that, if a reasonable attempt is made in the laboratory design to include the proper amount of mineral filler, this will minimize the amount of adjustment needed during field production. Quality Control of HMA AASHTO Standard Practice R 10, Definition of Terms Related to Quality and Statistics As Used in Highway Construction, defines quality assurance (QA) as "(1) all those planned and systematic actions necessary to provide confidence that a product or facility will perform satisfactorily in service; or (2) making sure the quality of a product is what it should be." For HMA production and pavement construction, the overall quality assurance process generally includes (1) quality control, (2) acceptance, and (3) independent assurance, which are defined in AASHTO R 10 as follows: Quality Control--The system used by a contractor to monitor, assess, and adjust their pro- duction or placement processes to ensure that the final product will meet the specified level of quality. Acceptance--The process whereby all factors used by the agency (i.e., sampling, testing, and inspection) are evaluated to determine the degree of compliance with contract requirements and to determine the corresponding value for a given product. Independent Assurance--Activities that are an unbiased and independent evaluation of all the sampling and testing (or inspection) procedures used in the quality assurance program. Obviously, quality control is essential to any business, including HMA production. Quality control is needed to ensure that the HMA produced at a given plant meets all of the specification requirements of the customer. An effective quality control program will also help lower costs and increase profits, because it will result in more efficient production. Many state highway agencies now use quality acceptance specifications when selecting and monitoring HMA producers and contractors for pavement construction. In a quality acceptance specification, the HMA producer is responsible for controlling HMA quality, while the highway agency is responsible for material acceptance. In HMA production, quality control refers to all actions and analyses performed by the plant to ensure that the material it produces is of good quality--practically speaking, that it meets all specifications and reasonable user expectations. Material acceptance is the procedure used by the state agency to determine if the HMA is acceptable, and if so, how much the plant will be paid for the material. The precise sampling and testing procedures used in product acceptance and the calculations to be followed in deciding on acceptance or rejection and payment factors are outlined in the agency's acceptance plan. Within the past 20 years, more and more states have implemented statistically based acceptance plans. These plans typically require random sampling of HMA, testing of multiple samples from a selected amount of HMA (often called a lot), and use of statistical calculations to determine product acceptance or rejection and payment amounts. It is critical to understand that these two operations--quality control and product acceptance--are completely separate and must not be confused. Although a state highway agency can and should review an HMA plant's quality control records to verify that it is in fact following a good plan, quality control test results should

OCR for page 211
212 A Manual for Design of Hot Mix Asphalt with Commentary not be used for acceptance. Ideally, the effectiveness of quality control operations at an HMA plant and the accuracy of product acceptance decisions made by a state agency should be regularly evaluated by an independent organization. This third part of a quality acceptance specification is independent assurance, as defined above. An important concept in both quality control and acceptance plans is that of quality charac- teristics. A quality characteristic is simply something about a product that is measured as an indication of its quality. Both HMA quality control and acceptance revolve around various quality characteristics: asphalt content, mineral filler content, aggregate gradation, in-place air void content, and so forth. Tests used to control the quality of HMA, and especially those used by agencies in making acceptance decisions, should be those most clearly related to pavement performance, but must also be fairly quick and easy to accurately measure. Both quality control and acceptance activities require an understanding of basic statistics-- calculation of average, standard deviation, mean, and other quantities. Quality control technicians must also be able to properly perform material sampling and testing and to then use such data to construct and use quality control charts to help keep a plant running smoothly and producing material that is within specifications. This has made the work of HMA technicians more difficult, but also more important to the success of plant operations. The sections below present an overview of the essential parts of quality control and acceptance plans. These discussions must be general, since the requirements for HMA quality control and acceptance vary significantly from state to state. HMA plant technicians should make sure they are familiar with specifications in their state and should take advantage of any training programs offered by their state highway agency or asphalt paving association on HMA quality assurance and related topics. Variability, Mean and Standard Deviation When engineers and technicians discuss a set of test results, the two most important charac- teristics of the data are the central tendency and variability. All real data sets exhibit variability, that is, the test results are not all exactly the same but vary within a certain range. The overall value of a data set is usually measured by calculating the mean value, often called the average. The mean is calculated by adding all the numbers in a data set and dividing by the total number of measurements. Variability in a set of numbers is usually measured by calculating the standard deviation: n ( xi - x ) 2 i =1 s= (12-1) n -1 Where s is the standard deviation xi is the value for the ith test is the mean value for all the tests x n is the total number of test results. Equation 12-1 is used for data sets with less than about 30 results; for larger data sets, n is used in place of (n - 1) in calculating standard deviation. Although Equation 12-1 may look complicated, most of the time standard deviation calculations are done using a calculator or spreadsheet. Calculators and spreadsheets will usually give the operator the choice of using n or (n - 1) in the standard deviation calculation; engineers and technicians need to make sure that (n - 1) is used when doing calculations for small data sets.

OCR for page 211
Field Adjustments and Quality Assurance of HMA Mixtures 213 Variability in HMA Production, Sampling, and Testing In designing and implementing a quality control program for HMA, it is critical to understand that asphalt concrete is, by nature, a highly variable material. Furthermore, the procedures we use to test aggregate, asphalt binder, and HMA are also variable. Even the procedures we use to sample materials at the plant and on the roadway cause variability in test data. Engineers and technicians must always consider these three sources of variability: materials, sampling, and testing. In a quality control program, the variability we should be most concerned with is the inherent variability in the material, for example, the range in the asphalt content for a particular HMA mix during a specific project. However, a substantial percentage of the variability in test data on pavement construction projects will be from test variability. Table 12-2 summarizes precision statements for different tests commonly used in HMA quality control plans. Precision statements are calculated using large data sets gathered from different laboratories on identical or nearly identical sets of specimens. They are usually given in terms of both standard deviation and the d2s precision. "d2s" stands for "difference, two standard deviations." In practical terms, the d2s precision represents the maximum expected difference between two test results on the same material. Precision statements also provide information on single-operator precision and multi- laboratory precision. Single-operator precision represents the variability expected for a given test when performed by the same technician in the same laboratory over and over again. The single- operator precision is useful in evaluating the quality of data being produced by a given technician. Laboratory managers can occasionally give technicians blind samples of the same materials, to determine if the test results are within the single-operator d2s limit. If not, it is likely the technician is performing the test improperly. The multilaboratory precision is probably more important to engineers and technicians involved in the paving industry. Multilaboratory precision represents the variability expected for a given test when it is performed repeatedly on the same material by a large number of different laboratories. The multilaboratory d2s is very useful when comparing test results produced by two different laboratories. For example, based on the data in Table 12-2, Table 12-2. Single operator and multilaboratory precision for test results commonly used in HMA quality control plans. Single Operator Multilaboratory Test Procedure Std. Dev. d2s Std. Dev. d2s Aggregate gradation, percent passing Coarse aggregate (CA)* 0.27 to 0.8 to 6.4 0.35 to 1.0 to 8.0 2.25 2.82 Fine aggregate (FA)* 0.14 to 0.4 o 2.4 0.23 to 0.6 to 4.0 0.83 1.41 Mineral Filler (in CA/ in FA) 0.10/0.15 0.28/0.43 0.22/0.29 0.62/0.82 Asphalt content, weight % Ignition oven 0.04 0.11 0.06 0.17 Quantitative extraction** 0.19 to 0.54 to 0.29 to 0.82 to 0.30 0.85 0.37 1.05 Maximum theoretical specific gravity 0.0040 0.011 0.0064 0.019 Bulk specific gravity, SSD 0.0124 0.035 0.0269 0.076 Bulk specific gravity, Paraffin-coated 0.028 0.079 0.034 0.095 Air void content, Vol. %*** 0.5 1.5 1.1 3.0 Effective asphalt content, Vol. %*** 0.3 0.9 0.6 1.6 Voids in mineral aggregate, Vol. %*** 0.5 1.5 1.1 3.1 Voids filled with asphalt, Vol. %*** 2.2 6.2 4.5 12.8 Dust/asphalt ratio, by weight*** 0.05 0.13 0.09 0.25 * Lower values are for very high and/or very low percent passing; higher values are for percent passing values close to 50%. ** Value depends on method used. *** Typical values, estimated from data on aggregate gradation, aggregate and mixture specific gravity and asphalt content using ignition oven. Values estimated using standard deviations for quantitative extraction vary slightly from these values.

OCR for page 211
214 A Manual for Design of Hot Mix Asphalt with Commentary Table 12-3. Effect air void content data on the same material from two different laboratories should not be considered of sample size on test suspect unless they differ by more than 3 to 4% for a single specimen. precision. The relatively large values for multilaboratory precision for some of the tests in Table 12-2 may Sample 1 Relative surprise and concern some engineers and technicians. Although test methods with better precision Size, n n Precision would improve quality control and acceptance procedures for asphalt concrete pavements, until 1 1.000 1.0 such methods are developed, the only way to improve the precision of these and other test methods 2 0.707 1.4 is to increase the number of tests performed on a given material. The number of tests performed 3 0.577 1.7 4 0.500 2.0 on a given HMA lot or section of pavement is usually called the sample size n; the precision of 5 0.447 2.2 any test performed on such a pavement increases with the square root of the sample size, n . 6 0.408 2.4 This is why many quality control and acceptance plans for HMA construction divide a lot into 7 0.378 2.6 8 0.354 2.8 subsections called sublots and require testing of one sample from each sublot. Such an approach 9 0.333 3.0 improves the precision of the measurements and makes the quality control and acceptance plan 16 0.250 4.0 more reliable. Table 12-3 gives values of 1/ n for different sample sizes n, along with relative 25 0.224 5.0 precision--a higher value of relative precision indicates an improved level of precision. Increasing the sample size from 1 to 4 doubles the precision of a measurement; increasing the sample size to 9 triples the precision. To increase the precision by a factor of 4, 16 replicate measurements are needed. To estimate the d2s precision when replicate measurements are made, or when more than one sample is taken from a given lot of material, the d2s precision from Table 12-2 (or from some other source) should be divided by the square root of the sample size, n , which is given as the "relative precision" in Table 12-3. As an example of how precision statements and sample size can be used in paving technology, consider a situation where a highway agency and contractor disagree on air void values. The agency's acceptance tests show an average air void content of 10.7% for a day's production of pavement, using an average of n = 4 measurements. The contractor's tests on companion samples gave an average of only 8.5%, using the same sample size. A penalty is applied if the air void content for a given lot exceeds 10%. The contractor believes the agency's tests are in error. Is the difference between these measurements surprising? Is further investigation needed to determine whether or not the air void content for this day's production is out of specification? According to Table 12-2, the d2s precision for air void content is 3 to 4%. For a sample size of n = 4, the relative precision would be cut in half, reducing the d2s precision to 1.5 to 2.0%. The difference in the measurements is 10.7 - 8.5 = 2.2%, which is greater than the calculated d2s precision. Therefore, the difference in these test values is too large and should be investigated. Control Charts Control charts are one of the most important parts of a good quality control program. Control charts are simply a way of using a graph to show how important test results are changing over time at a given HMA plant. An experienced technician or engineer can look at a set of control charts and determine if a plant is operating smoothly and producing acceptable material or if there is a production problem that requires investigation and perhaps adjustment. Control charts are not just used at HMA plants, but form an essential part of quality control plans in most manufacturing industries. Several types of control charts are commonly used in HMA production. One of the simplest is made by plotting test results as a function of time. Such a control chart is shown in Figure 12-1, which shows asphalt content as a function of date and time. In this case, the target asphalt content is 4.8%. The specification requires that individual asphalt content measurements fall within 0.7% of the target. To make the control chart more useful, both the target and the upper and lower specification limits are shown on this control chart. Note that on or about June 24, many of the

OCR for page 211
Field Adjustments and Quality Assurance of HMA Mixtures 215 6.8 Asphalt Content, Wt. % upper specification limit target 4.8 lower specification limit 2.8 1 2 1 2 1 2 1 1 2 1 2 0 M2 1 21 M2 1 21 M2 1 2 2 1 2 1 2 1 2 1 2 M 6/ AM 6/ PM 6/ PM 6/ AM 6/ AM 6/ PM 6/ AM 6/ AM 6/ PM 6/ PM 6/ AM 6/ PM 6/ PM 6/ AM 6/ PM 6/ PM 6/ AM 6/ AM 6/ PM 6/ PM 6/ AM M -A -P -P -A -A - - - - - - - - - - - - - - - - - - - - - 5 14 14 14 15 15 15 16 16 16 16 20 21 21 24 24 24 27 27 27 27 28 28 /0 6/ 6/ 6/ 6/ 14 6/ Date - Time Figure 12-1. Simple control chart showing single measurements of asphalt content. asphalt contents are near the specification minimum, and on the morning of June 27, one of the measurements is below the specification minimum. Figure 12-1 is useful, but has several drawbacks--it only shows single measurements, and the high level of variation among single test results makes overall trends hard to see. Figure 12-2 is a control chart of the same data, but here running averages are plotted instead of single measurements. The specification requirement shown in this plot is that the average asphalt content for the lot must be within 0.4% of the target. Because the specification in this case is based on the average of n = 5 samples, the running average was calculated using the five latest asphalt content tests. Because this method reduces much of the test variability seen in Figure 12-1, Figure 12-2 is much smoother. It is clear from Figure 12-2 that the average asphalt content is slowly decreasing. Figure 12-2 is also useful because it gives the technician an idea of whether or not lot averages will meet the specification requirement. In actual practice, lots in acceptance plans for pavement construction are most often based on 1 day's production. For this reason, Figure 12-2 gives only an approximate indication of whether or not the lot average asphalt content will meet specification requirements. A better indication of whether or not the requirement for lot average asphalt content is being met can be gained by plotting average asphalt content as a function of the date of production, as shown in Figure 12-3. This plot shows the steady decrease in asphalt content seen in Figure 12-2, but also suggests that production for the last 2 days is out of specification. Although Figure 12-3 will better reflect the results of acceptance based on daily production, it should be remembered that such control charts are based on the plants quality control data--acceptance test results may be different, because of 5.8 Avg. AC, Wt. % upper specification limit target 4.8 lower specification limit 3.8 1 2 1 2 1 2 1 1 2 1 2 1 2 1 2 1 2 2 1 2 1 2 1 2 1 2 M 6/ AM 6/ PM 6/ PM 6/ AM 6/ AM 6/ PM 6/ AM 6/ AM 6/ PM 6/ PM 6/ PM 6/ PM 6/ AM 6/ AM 6/ PM 6/ PM 6/ AM 6/ PM 6/ PM 6/ AM 6/ AM 6/ PM 6/ PM 6/ AM M -A -A - - - - - - - - - - - - - - - - - - - - - - - - 5 14 14 14 15 15 15 16 16 16 16 20 20 21 21 21 21 24 24 24 27 27 27 27 28 28 /0 6/ 14 6/ Date - Time Figure 12-2. Control chart for asphalt content using running average of five measurements.

OCR for page 211
216 A Manual for Design of Hot Mix Asphalt with Commentary 5.8 Avg. AC, Wt. % upper specification limit target 4.8 lower specification limit 3.8 5 15 16 20 21 24 27 /0 6/ 6/ 6/ 6/ 6/ 6/ 14 6/ Date Figure 12-3. Control chart for average asphalt content, as a function of production day. differences in location of samples and sampling methods and lab-to-lab variability in test results. Also, using daily averages to construct control charts will mean that the charts cannot be constructed until test data for each day are complete and have been tabulated. This will delay creation of the control chart and identification and correction of any problems that occur. Plotting the specification limits on control charts can be useful, but HMA production can usually be better controlled by using statistical control limits. Sometimes these limits are based on three times the average standard deviation. Because the Greek symbol ("sigma") is used to represent the actual standard deviation of a large group, these kinds of control limits are sometimes called three sigma (3) limits. These limits are used because the chances of a given test result falling outside these boundaries are very small--about one in one thousand--so a test result near or outside of these control limits should be investigated immediately. A similar chart often used in quality control for HMA plants uses the average range (R or R-bar) instead of the standard deviation to calculate control limits; the target value is the overall average, rather than the center of the specification limits. Figure 12-4 is an example of such a chart, constructed for the daily average percent passing the 0.075-mm sieve. In this case, the overall average range for percent passing the 0.075-mm sieve for this plant was estimated to be 1.15%, and the overall average was 4.2%. The control limits are then calculated using the following formulas: UCL = X + ( A2 R ) (12-2) LCL = X - ( A2 R ) (12-3) 6.0 0.075-mm Sieve, Wt. % Daily Avg. Passing upper control limit 5.0 4.0 3.0 lower control limit 2.0 9/8 9/9 9/12 9/13 9/14 9/16 9/19 9/20 9/21 Date Figure 12-4. Statistical control chart for average percent passing the no. 200 sieve.

OCR for page 211
Field Adjustments and Quality Assurance of HMA Mixtures 217 Table 12-4. Factors for computing control limits for control charts. Sample Size n A2 D3 D4 2 1.88 0.00 3.27 3 1.02 0.00 2.58 4 0.73 0.00 2.28 5 0.58 0.00 2.12 6 0.48 0.00 2.00 7 0.42 0.08 1.92 where UCL = upper control limit LCL = lower control limit _ = overall average or X-bar = 4.2% for Figure 12-4 X R = overall range or R-bar = 1.35% for Figure 12-4 A2 = a factor that depends on sample size n (given in Table 12-4); in this example n = 5 and A2 = 0.58 The average mineral filler contents in Figure 12-4 are within the control limits and close to the target until September 19. At this point, the values become much more variable, going below the lower control limit on September 19 and 21, and near the upper control limit on September 20. Clearly, there is a serious problem with variability in the mineral filler at this plant starting on September 19, and the situation should be investigated to determine the source of the problem. Because specification limits for average percent passing the 0.075-mm sieve are often 2.0%, the data shown in Figure 12-4 would probably be within the specification and would not represent a payment penalty. If the specification limits of 2.2 to 6.2% were used as the limits in Figure 12-4, it might not be clear that there is a problem in the plant operation until some data exceed the specification limits, incurring a payment penalty. The advantage of using statistical control limits is that potential problems in production can often be identified and corrected before material exceeds specification limits. An important question in the construction of statistical control charts is what values to use for the overall average and range. In many cases, a well-run plant will continuously collect and analyze information on variability in asphalt content, aggregate gradation, laboratory air void content, and other important aspects of HMA production. The plant operator or technician in charge of quality control should then have values for overall average and range. If not, Equation 12-1 can be used to calculate overall standard deviation for a given property for a given mix type from the available data. However, for reasonably accurate estimates of overall average and range, data from at least 30 production days are needed. Production data can be used once data from about 20 days are available, although the results will not be completely reliable. When calculating data for use in control charts, it should be emphasized that production vari- ability will change over time for several reasons: employee turnover, equipment wear, changes in plant layout, improvements in employee training, and so forth. Therefore, overall average and range values should be recalculated regularly, using the last 30 to 50 data points available for a given test. Typical overall standard deviations for different HMA test properties are listed in Tables 12-5 and 12-6; these values have been gathered and reported in research studies listed at the end of this chapter. Expected values for R (overall range) can be estimated from values given in this table by dividing the standard deviation by the factor A2 found in Table 12-4; remember

OCR for page 211
218 A Manual for Design of Hot Mix Asphalt with Commentary Table 12-5. Typical overall standard deviation values for aggregate gradation. Typical Range for Sieve Size Overall Standard Deviation 19 mm 1.5 to 4.5% 12.5 mm 2.5 to 5.0% 9.5 mm 2.5 to 5.0% 4.75 mm 2.5 to 5.0% 2.36 mm 2.5 to 4.0% 1.18 mm 2.5 to 4.0% 0.60 mm 2.0 to 3.5% 0.30 mm 1.0 to 2.0% 0.15 mm 1.0 to 2.0% 0.075 mm 0.6 to 1.0% Table 12-6. Typical overall standard deviation values for asphalt content, air void content, VMA and VFA. Typical Range of Value for Overall Property Standard Deviation Asphalt content 0.15 to 0.30% Air void content, from field cores 1.3 to 1.5% Laboratory air void content 0.9% VMA 0.9% VFA 4.0% that the value of A2 and the resulting estimate of R will depend on the sample size n--the larger the sample size, the larger the value of R for a given standard deviation. All of the control charts shown above are plots of measured values or of averages of measured values. In another type of control chart, values of the range (as determined in each lot's or day's production) are plotted as a function of time of sampling. In Figure 12-5, the same data used in Figure 12-4 were used to calculate range values for each production day. The control limit in this case is calculated by multiplying R (1.35%) by a factor that depends on sample size: LCL = D3 R = 0.0 1.35 = 0.0 (12-4) UCL = D4 R = 2.12 1.35 = 2.86 (12-5) 6.0 Daily Range Passing 5.0 No. 200, Wt. % 4.0 upper control limit 3.0 2.0 1.0 0.0 9/8 9/9 9/12 9/13 9/14 9/16 9/19 9/20 9/21 Date Figure 12-5. Range control chart for percent passing the no. 200 sieve, plotted by date.

OCR for page 211
Field Adjustments and Quality Assurance of HMA Mixtures 219 where LCL = lower control limit = 0.0 for Figure 12-5 UCL = upper control limit = 2.86% for Figure 12-5 D3, D4 = factors for computing control limits for standard deviation control charts; see Table 12-4 R = overall range = 1.35% for Figure 12-5 Figure 12-5 makes it clear what the problem is with the mineral filler data--the variability suddenly increases starting on September 16. After this date, most of the standard deviations are above the upper control limit, a situation that must be investigated and corrected. Note that there is no lower control limit in Figure 12-5. This is because the factor D3 for calculating the LCL (see Table 12-4) is zero for a sample size of 6 or less and need not be plotted. Even for n = 7, the value of D3 is 0.08, which will still result in very low values for the LCL relative to the UCL. Control charts are an important part of quality control at HMA plants. They are useful tools for adjusting plant production to help ensure that specifications are met, work stoppages minimized, and payment penalties avoided. The number and types of charts used at a given HMA plant will depend on the specifications that the plant must usually meet and on the preferences of the technicians and engineers responsible for running the plant. As a minimum, it is suggested that control charts should be kept for asphalt content, percent passing the No. 200 sieve, percent passing the No. 8 sieve, air void content of laboratory-compacted specimens, and VMA of laboratory-compacted specimens. Technicians responsible for plant quality control may also be required to keep control charts for in-place air void content and pavement thick- ness. For each set of test data, it is suggested that plots be made for individual measurements, running averages, running standard deviation, daily average, and daily standard deviation. The number used for calculating running averages and standard deviations should be equal to the number of sublots generally used in the local agency's acceptance plan--usually n = 4 or n = 5. There are many rules for interpreting statistical control charts. A few simple rules for when to investigate a potential problem will help technicians and engineers use these tools effectively: One or more points outside UCL or LCL Seven or more points on one side of target for x charts A gradual increase or decrease in the value of x A gradual increase in standard deviation or standard deviation A sudden shift in x Daily or weekly variations in x When one of these potential problems is seen in a control chart, the following procedure is suggested for investigating the reason for the unusual data: 1. Check to make sure that the data were correctly recorded. 2. Check the calculation of the test data. 3. Interview the technician or engineer responsible for taking the sample, to determine if there was anything unusual about the sample. 4. Interview the technician or engineer responsible for performing the test, to determine if anything unusual occurred when the test was performed. 5. If the data appear to be reliable--if there were no errors in sampling, testing, calculating, and recording the data--investigate those parts of the HMA plant or production process that could affect the test result. This might include valves, meters, scales, screens, aggregate stockpiles, and so forth.

OCR for page 211
220 A Manual for Design of Hot Mix Asphalt with Commentary 6. If a reason for the unusual data is found, it is called an "assignable cause." The problem should be corrected as soon as possible. 7. If no problems are found in the plant or process, the unusual test data are determined to be the result of "chance cause," meaning that it was likely the result of normal variability in materials, sampling, or testing. Technicians and engineers responsible for controlling an HMA plant should be careful not to over-control. Over-control occurs when adjustments are made in the HMA plant that are not needed. This might sound harmless, but over-control is a serious problem that can lead to HMA being out of specification, resulting in work delays and penalties. Consider the following example. An inexperienced plant operator who does not fully understand control charts sees that a few values for aggregate passing the No. 8 sieve are a few percent above the target, so the operator decreases the fine aggregate in the aggregate blend by 2%. The next day, the operator notes that the fine aggregate is now running about 3% low, so the operator increases the fine aggregate by the same amount. By the end of the day, the operator notes that the fine aggregate is 5% high, and the operator now decreases the setting by 4%. The next morning, the new plant operator realizes that the plant has produced nearly a full day's production that is out of specification because of highly variable aggregate gradation, including the mineral filler content. In reality, the variability that the operator first saw was normal, and no adjustment in the process was needed. When the operator made the first "tweak" in the fine aggregate, this operator was only making the situation worse, by adding an additional source of variability to the process. Each additional adjustment only made the situation worse. Adjustments in the plant process should not be made unless truly unusual data are observed and, as discussed above, an assignable cause is found. A second important aspect of quality assurance involves continuous quality improvement. As discussed earlier in this chapter, a certain amount of variability in quality control data is normal-- it is the result of variation in sampling, test procedures, and HMA plant operations. However, this does not mean that plant personnel need to be satisfied with a given level of variability. Continuous quality improvement is the process of identifying and reducing or eliminating sources of variability. This might mean training laboratory technicians to reduce variability in sampling and testing. Purchasing new or improved test equipment can also decrease testing variability. High variability in aggregate gradation might be caused by improper stockpiling--proper training of plant personnel responsible for aggregate handling will result in more uniform stockpiles and reduced variability in aggregate gradation. If mechanical parts in the plant are wearing out regularly, causing spikes in production variability when they fail, they can be replaced regularly before they fail. Although a good quality control program might seem an unnecessary cost, it is now widely accepted in many industries that proper quality control will save more money than it costs. Acceptance Testing During HMA Production As mentioned above, acceptance decisions in HMA production are the responsibility of the highway agency. In order to keep costs low and to ensure that agency decisions can be made quickly, acceptance testing is usually more limited than quality control testing performed by contractors. Tests included in acceptance plans might include asphalt content, percent of aggregate passing the 0.075-mm sieve, percent of aggregate passing the 2.36-mm sieve (or some other intermediate sieve size), in-place air void content, pavement smoothness, and pavement thickness. Pavement smoothness is usually measured using a profilometer or similar device. Sampling for in-place air void content and pavement thickness must be done from the finished pavement, either by sawing slabs or removing cores from the pavement. HMA samples for asphalt content and aggregate gradation can be cores taken from the finished pavement, loose mix taken from behind the paver,

OCR for page 211
Field Adjustments and Quality Assurance of HMA Mixtures 221 or loose mix taken from the truck at the plant site. It is essential that the sampling plan clearly describe the location of the sample, the size of the sample, and the procedure to be used in taking the sample. Samples from the actual pavement--either cores or loose mix taken from behind the paver--are preferred for acceptance tests because they more closely represent the final product than do samples taken from the truck or other locations within the plant. HMA plant technicians and engineers should study their state agency's acceptance plans, so that they can understand where problems might occur that might affect test results and acceptance decisions. For example, inexperienced inspectors might use poor techniques in sampling HMA, leading to highly variable test results and rejection of good-quality HMA. Acceptance sampling and testing may be done by employees of the highway agency or by an independent engineering firm or testing laboratory. Sometimes, because of reductions in state government staffing or budgets, contractors may be required to perform acceptance sampling and testing. As a general principle, such acceptance tests should be kept separate from quality control tests. Acceptance testing, regardless of who is responsible for performing it, should be verified regularly by an independent third party, through inspections, interviews, and comparison of test data on samples taken from the same pavement section. Sampling Materials for Quality Control and Acceptance Testing It is critical in both quality control and acceptance testing that the samples tested are repre- sentative of what is being produced or of the materials used in production. AASHTO publishes three standard methods for sampling aggregate, asphalt binders, and HMA: T 2, Sampling of Aggregates T 40, Sampling Bituminous Materials T 168, Sampling Bituminous Paving Mixtures Engineers and technicians responsible for collecting samples for quality control and acceptance testing should follow these procedures or appropriate procedures provided by their state highway agency. Often large samples of material taken during production must be reduced in size for testing purposes. Again, appropriate procedures for sample splitting, as published by AASHTO or the local state highway agency, should be followed. This is especially important for aggregates, where segregation of samples during handling and transport is a common problem. AASHTO T 248, Reducing Samples of Aggregate to Testing Size, describes appropriate procedures for reducing large samples of aggregate for testing purposes. Although there is no AASHTO standard for reducing large samples of binder to testing size, careless handling of binder in the laboratory can also cause problems during mix design, quality control testing, and acceptance testing. The length of time and number of times asphalt binder is heated in the laboratory should be kept to a minimum. This means that if a large sample of binder, such as a quart or gallon, is to be divided for testing and mix design, the sample should be heated just until fluid and then poured into a range of smaller containers. These smaller containers are then used for testing and mix design. Large containers of binder should not be kept in an oven for an extended period of time and should not be repeatedly reheated for testing and mix design. Such extended or repeated heating will harden the binder, yielding erroneous binder test results and potential inconsistencies in HMA mix designs. As with asphalt binder (and for similar reasons), HMA mixture samples should not be heated repeatedly or for extended periods of time. Also, when HMA samples are split for testing, care should be taken to make sure that asphalt binder and fines are not lost on containers and laboratory tools. Some HMA mixtures are prone to segregation, and special care is needed in handling such mixtures. Many apparent problems in quality control and acceptance testing during HMA production are the result of improper sampling practices or segregation during sample handling. Following

OCR for page 211
222 A Manual for Design of Hot Mix Asphalt with Commentary the appropriate AASHTO procedures, or those provided by the local highway agency, will minimize such problems. When specific procedures for sampling and sample size reduction are not available, reasonable care and common sense will go far to eliminate potential problems. Equally important to proper sampling technique is developing and following an appropriate sampling plan. A sampling plan describes when and where to take samples for testing and also what type of sample should be taken and how large the sample should be. Samples should be taken randomly, meaning that the exact time or location should vary in unpredictable ways. This prevents production personnel from anticipating when and where samples will be taken and changing their practices to make certain that the results of quality control or acceptance test- ing are favorable. Random number tables are often used to generate the times and locations of sampling in HMA quality control and acceptance sampling plans. Many HMA acceptance plans are stratified random sampling plans. Such plans divide an amount of material to be tested into lots and sublots. Samples are taken from each sublot within a lot and tested, and the results are used to calculate average values, standard deviation, and other statistics for the entire lot. These statistics are then used by the agency to decide whether the lot should be accepted or rejected or if the lot should be accepted at reduced payment. As an example, consider the main features of a typical stratified random sampling plan for HMA acceptance: HMA samples weighing between 2 and 3 kg will be taken behind the paver, before compaction. Samples will be stored in clean, unlined, tightly sealed metal cans. For purposes of sampling, a lot is normally defined as 1,000 tonnes of hot mix. Each such lot will be divided into five equal sublots of 200 tonnes each. One HMA sample will be taken from each sublot. The location of this sample, both along and across the roadway, will be determined using a set of two random numbers, one representing the location of the sample along the roadway, and the other representing the location across the pavement. Samples should not be taken within 2 feet of obstructions to paving. The resulting sampling of a typical lot might look something like that sketched in Figure 12-6. The coordinates (x, y) for each sample are determined from a random number table. The example given above is greatly simplified; real specifications must provide detailed directions for a wide range of situations, so that there is no confusion about when and where to take samples for acceptance testing. For example, poor weather may result in lots smaller than the normal 1,000 tonnes described above; an effective specification must explain clearly how to deal with such situations. Technicians and engineers responsible for sampling should make sure they are familiar with the specification that describes the sampling plan that they should be following and y x sublot 1 sublot 2 sublot 3 sublot 4 sublot 5 lot direction of paving Figure 12-6. Example of a stratified random sampling plan for HMA.

OCR for page 211
Field Adjustments and Quality Assurance of HMA Mixtures 223 should also understand the importance of obtaining representative samples for both quality control and acceptance testing. AASHTO provides detailed guidance for developing acceptance plans in R 9, Acceptance Sampling Plans for Highway Construction. Quality Control Plans AASHTO R 10, Definition of Terms for Specification and Procedures, gives the following definition of a quality control plan: A detailed description of the type and frequency of inspection, sampling, and testing deemed necessary to measure and control the various properties governed by agency specifications. This document is submitted to the agency for approval by the contractor during the preconstruction conference. Although quality control plans are meant to be documents used by the HMA supplier or contractor to help ensure that their product meets the customer's needs and will thus be accepted at full payment, many highway agencies will require the supplier to submit a quality control plan for approval. This gives the highway agency additional confidence that the HMA supplier will be able to meet the required specification. The highway agency may list the items to be included in the quality control plan. A typical quality control plan for an HMA supplier might include the following items, as a minimum: 1. Quality Control Organization Chart 1.1. Names of personnel responsible for quality control 1.2. Area of responsibility of each individual 1.3. List of outside agencies, such as testing laboratories, involved in quality control and a description of services provided by each 2. Testing Plan with Action Points 2.1. List of all tests to be performed 2.2. Frequency of testing 2.3. List of action points for initiating corrective procedures 2.4. Recording method to be used to document corrective procedures 3. Materials Storage and Handling 3.1. Aggregate and RAP stockpiles 3.2. Cold-feed systems for aggregates and/or RAP 3.3. Additives or modifiers 3.4. Asphalt binder and other liquid storage tanks 3.5. Surge and storage silos for HMA 3.6. All measuring and conveying devices, including calibration procedures for each 3.7. Description of procedure for loading haul vehicles Engineers and technicians responsible for developing quality control plans for HMA production should follow guidelines provided by their local highway agencies. They should also keep in mind that, to be effective, a quality control plan should be as simple as possible, while addressing all important items and activities that affect the quality of HMA produced at the plant. Bibliography AASHTO Standards R 9, Acceptance Sampling Plans for Highway Construction R 10, Definition of Terms for Specification and Procedures T 2, Sampling of Aggregates T 40, Sampling Bituminous Materials T 168, Sampling Bituminous Paving Mixtures

OCR for page 211
224 A Manual for Design of Hot Mix Asphalt with Commentary Other Standards ASTM D 3665, Random Sampling of Construction Materials "Section 409--Superpave Mixture Design, Standard and RPS Construction of Plant-Mixed HMA Courses" in Publication 408, Harrisburg, PA: The Pennsylvania Department of Transportation, 2007. Other Publications Burati, J. L., Jr., and C. S. Hughes (1990) Highway Materials Engineering, Module I: Materials Control and Acceptance--Quality Assurance, National Highway Institute Course No. 13123, February. Hot-Mix Asphalt Paving Handbook. (1991) AC150/5370-14, Appendix I; UN-13 (CEMP-ET), J. Sherocman, Consultant, AASHTO, FAA, FHWA, NAPA, USACE, American Public Works Association, National Association of County Engineers, July 31, 218 pp. NCAT (1996) Hot-Mix Asphalt Materials, Mixture Design and Construction, NAPA, Lanham, MD, 585 pp.