Tear-offs in Asphalt Concrete

This is the second of a two part series on a study of recycled asphalt shingles performed by the Virginia Transportation Research Council. This section looks
at the results of field tests performed in the state.
By G.W. Maupin Jr., PE

Superior Paving Corp., Stevensburg, Va.—The BM-25.0 base mix was sampled twice as it was placed on two VDOT paving projects. The first sample was taken by the contractor’s personnel on April 28, 2009, for a widening paving project on Route 230. Only volumetric properties and a binder recovery/grading were obtained on this sample. The same base mix was sampled on May 14, 2009, as it was being used to pave a section on Route 15 near a shopping center north of Culpeper, Va. For both projects, the mix was produced through a double drum plant at an average production temperature of 295˚F to 300˚F. The ground shingles were added through a RAP bin. Paving took place normally with no problems.

W-L Construction & Paving Inc., Strasburg, Va.—The BM-25.0 base mix was sampled and placed on a small subdivision street on June 22, 2009. The mix was produced through a batch plant at an average temperature of 300˚F. The shingles were pre-blended in a 50:50 blend with No. 10 aggregate and stockpiled before being fed into the drum through the RAP collar in proper portions resulting in 4% shingles in the asphalt mix. Paving took place normally with no problems, although the mix did exhibit a tender zone that necessitated a delay of rolling during compaction.
Branscome Inc., Richmond, Va.—The SM-12.5 surface mix was sampled and placed on Route 460 west of Petersburg, Va., on July 30, 2009. The mix was produced with a double drum plant at an average temperature of 300˚F. The mix contained a combination of recycled materials consisting of 18% RAP and 2% shingles. The two materials were added concurrently to the RAP conveyor belt in the proper proportions and entered the plant drum through the conventional RAP collar. No problems were encountered with the paving process.

W-L Construction & Paving Inc., Clear Brook, Va.—The SM-12.5 surface mix was sampled and placed on the shoulder of Route 522 south of Winchester, Va., on October 7, 2009. The HMA and WMA were produced through the same double drum plant at average temperatures of 300˚F and 270˚F, respectively. The WMA was produced using the green plant foaming system that used a very small amount of water to cause a foaming action of the asphalt. The shingles were pre-blended in a 50:50 blend with No. 10 aggregate before being entered into the drum through the RAP collar. The target shingle content was 5%. Paving proceeded with no observed problems.

Laboratory Tests
Volumetric Properties and Ignition Furnace Results—Volumetric properties and gradation results for the field samples from each project are shown in Tables 2 and 3, respectively. The volumetric properties, air voids and VFA were all within acceptable production limits. VMA was greater or very close to the minimum design values. The property values for the samples from Superior Paving projects were very close, which indicates consistency in the product from day to day. Similarly, the asphalt contents and gradations representing fine sieve sizes that might be affected by shingles were very close for the samples from Superior Paving, further verifying consistency in the product. The asphalt contents of single samples for all projects were well within production limits, i.e., +0.3% for the average of four samples.

W-L Construction & Paving Inc. indicated that the shingles were pre-blended with the No. 10 aggregate in a 50:50 ratio. Samples of the blended material were tested for asphalt binder content with the ignition furnace method for their base mix and surface mix projects. With the ignition furnace it was determined that the blend contained shingle/No. 10 aggregate ratios of 33:67 and 37:65 for the base mix and surface mix projects, respectively. However, the contractor apparently adjusted the amount of the blended material entering the plant to yield the proper mix binder content, as evidenced by the production mix binder contents being very close to the target job mix values.

Rut Tests
The rut test results are shown in Table 4. According to VDOT’s maximum allowable limits, the test results indicated that the mixes would be satisfactory for heavy traffic situations from a rutting resistance standpoint. In addition, the rut depths were comparable to those reported for conventional Virginia D surface mixes containing PG 70-22 binder that were tested in an earlier high-RAP study.

Fatigue Tests
The fatigue regression constants and endurance limits as described previously are shown in Table 5. The fatigue regression log-log plots are shown in Figure 1. A previous high-RAP study tested eight mixes containing 21% to 30% RAP that yielded endurance limits ranging from 83 to 130 microstrain.9 Those mixes contained considerable RAP and are used routinely by VDOT. The fatigue endurance limits of the mixes containing shingles were within the range of values for endurance limits of mixes currently allowed by VDOT.

Binder Recoveries and Grading
The performance gradings of the binders recovered from field samples of the projects are listed in Table 6. In reference to the virgin PG 64-22 binder, the high-temperature grade was increased one grade on three of the projects, two grades on two of the projects, and three grades on one of the projects. The low-temperature grade deteriorated one grade on five of the cases and stayed the same in the sixth case. However, in all cases, the binders were a -16 low temperature grade or better, which passed VDOT specifications. Bonaquist reported at the 4th Asphalt Shingle Recycling Forum in Chicago that the addition of 25% roofing shingle binder improves the high-temperature grade two levels and makes the low-temperature grade one grade poorer, further supporting the results of the present study. There was no apparent difference in the properties of the binder recovered from the HMA and WMA samples for the W-L Construction & Paving project on Route 522. Perhaps the temperature difference of 30˚F for the production of these mixes was not sufficient to cause stiffening of the binder.

Additional Laboratory Testing
Shingle Material Properties—Samples of shingles were subjected to binder removal by solvent extraction and by the ignition furnace, which is the generally accepted method for asphalt content determination. The solvent extraction would be considered to yield the “true” binder content, whereas the ignition furnace also burns some foreign material and possibly aggregate. The purpose was to determine an ignition furnace correction factor that should be applied when quality control/quality acceptance tests are performed.
Two samples of ground shingles were tested by solvent extraction, yielding 24.3% binder; companion samples yielded 29.2% binder by the ignition furnace test. In other words, the ignition furnace test indicated that the shingles contained about 5% more binder than they actually contained. A similar determination by South Carolina found that the difference between extraction and ignition testing was 2% for its shingles. Assuming a correction of 5% for shingles, the correction for an ignition furnace test that should be applied to the binder content of mix containing shingles would be:

[(% shingles) x (% difference between binder determined by extraction and ignition furnace)] /100

Therefore, an approximate 0.05% shingle correction factor for each 1% of shingles would need to be applied to the ignition furnace results for binder content of a mix. A mix containing 4% shingles would need to have a 0.2% correction applied in addition to the normal aggregate correction. This factor was determined only from a limited sampling and could vary somewhat for other sources of shingles.

Binder Blending for Warm Mix
An SM-9.5D surface mix containing PG 70-22 binder was tested in indirect tension. The tests were performed on specimens containing shingle contents ranging from 0to 5% that had been mixed at two temperatures: 250˚F and 300˚F. The specimens were mixed at 300˚F to simulate HMA and at 250˚F to simulate WMA. The idea was to determine how mix stiffness was affected by the mixing  temperature. If the shingle binder and virgin binder combined to the same degree at both mixing temperatures, the mix stiffness should have displayed the same differential increase at both temperatures as shingle content was increased.

Table 7 lists the strengths and Figure 2 shows graphically the strength difference attributable to inadequate blending of the virgin and shingle binders. The 300˚F curve was shifted down to the 250˚F curve to eliminate the additional aging effects. The strength difference between the shifted curve and 250˚F curve can then be attributed to inadequate blending. With 5% shingles, the strength of the WMA mix increased only about 50% as much as that of the HMA mix [(190-155) / (245-183)] = 0.56. It is difficult to duplicate field conditions in the laboratory, and perhaps better binder blending would normally occur in a hot-mix plant than in a laboratory experiment because of more vigorous mixing.

The construction experience of contractors with tear-off shingles was satisfactory with no problems. The volumetric properties of mix from each project were within VDOT specifications. Rut tests indicated that the mixes containing shingles can be used in situations involving heavy traffic where rutting is a concern. Fatigue test results indicated that fatigue durability is comparable to that being provided by current mixes. Binder recoveries indicated an improvement of high-temperature grading, and the low-temperature grading was within VDOT specifications. Limited indirect tensile testing indicated that blending of the virgin and shingle binder was less for WMA than for HMA. This is an area for possible research.

MoDOT allows up to 30% of the shingle binder to be used without a change in the virgin binder grade, and it has used a considerable amount of shingles in its hot mix since 2005 with satisfactory performance. MoDOT used approximately 53,000 tons of shingles in 2009, of which about 90% was tear-off shingles. Considering Missouri’s good pavement performance, the current VDOT special provision allowing a maximum of 25% aged shingle binder versus MoDOT’s 30% shingle binder with no change in virgin binder grade should be conservative.

•    The use of tear-off roofing shingles in asphalt mixes should produce satisfactory mixes.
•    Routine mix properties during construction should be satisfactory.
•    Rutting resistance complies with the requirements for VDOT mixes used in heavy traffic.
•    Shingle mixes perform as well as conventional high-RAP mixes.
•    The high-temperature grading of binders was improved, and the low-temperature grading was within VDOT specifications.
•    The ignition furnace correction factor should be based not only on aggregate correction but also on the combustibility of non-asphalt materials contained in the shingle waste.

1. VDOT’s Materials Division should develop a modified specification stating that recycled shingles must be certified to be free of asbestos-containing
material as defined by the National Emission Standards for Hazardous Air Pollutants (i.e., a material containing less than 1% asbestos) and that the source of recycled shingles must be residential homes with no more than four units per structure. Such shingles have very little chance of containing asbestos. Such specifications would possibly minimize the amount of required asbestos testing but would necessitate certification by the recycler concerning the source of the product.
2. With the changes noted in Recommendation 1, VDOT’s Materials Division should continue the use of the current special provision allowing tear-off shingle waste for general usage.

Costs and Benefit Analysis
A representative of Oldcastle Materials, Inc., of Atlanta, Ga., reported at the 4th Asphalt Shingle Recycling Forum held in 2009 that the cost savings per ton of plant mix is approximately $3 to $5 when reclaimed shingles are used. The source of the shingles used in HMA in Virginia is Asphalt Roof Recycling Co., Mt. Airy, Md. Assuming an average cost of virgin binder for VDOT’s 2009 construction season of approximately $400 per ton and the FOB price of reclaimed shingles from Mt. Airy, a savings of approximately $3.40 per ton of HMA would be realized with the use of reclaimed shingles. If the transportation costs of hauling the shingles to the asphalt plants of the contractors involved in this study is also considered, the savings would be reduced by $0.40 to $1 per ton. If the process is approved by VDOT for general usage, shingle recyclers will be located near the facilities of asphalt contractors, resulting in lower shingle costs. If a savings of $3 per ton could have been realized on one-half of the 400,000 tons of HMA produced last year for VDOT, VDOT would have saved approximately $600,000. Potential cost savings is further evidenced by the fact that contractors were willing to try the process with no additional compensation from VDOT and some contractors continued to use the shingles on private work.

As already mentioned, a considerable amount of shingles in Virginia is currently being disposed of in waste facilities. The use of recycled shingles would not only provide cost savings for VDOT but would also be an environmental plus for Virginians since waste material that would otherwise go to a landfill would be used. The use of shingle waste in asphalt concrete would also create possible business opportunities in recycling and processing the shingles. 

G.W. “Bill” Maupin recently retired as principal research scientist at the Virginia Research Council. He can be reached at [email protected]