National Academies Press: OpenBook
« Previous: Chapter 2 - Research Approach
Page 25
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 25
Page 26
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 26
Page 27
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 27
Page 28
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 28
Page 29
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 29
Page 30
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 30
Page 31
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 31
Page 32
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 32
Page 33
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 33
Page 34
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 34
Page 35
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 35
Page 36
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 36
Page 37
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 37
Page 38
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 38
Page 39
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 39
Page 40
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 40
Page 41
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 41
Page 42
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 42
Page 43
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 43
Page 44
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 44
Page 45
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 45
Page 46
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 46
Page 47
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 47
Page 48
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 48
Page 49
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 49
Page 50
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 50
Page 51
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 51
Page 52
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 52
Page 53
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 53
Page 54
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 54
Page 55
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 55
Page 56
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 56
Page 57
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 57
Page 58
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 58
Page 59
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 59
Page 60
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 60
Page 61
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 61
Page 62
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 62
Page 63
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 63
Page 64
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 64
Page 65
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 65
Page 66
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 66
Page 67
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 67
Page 68
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 68
Page 69
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 69
Page 70
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 70
Page 71
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 71
Page 72
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 72
Page 73
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 73
Page 74
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 74
Page 75
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 75
Page 76
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 76
Page 77
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 77
Page 78
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 78
Page 79
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 79
Page 80
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 80
Page 81
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 81
Page 82
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 82
Page 83
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 83
Page 84
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 84
Page 85
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 85
Page 86
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 86
Page 87
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 87
Page 88
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 88
Page 89
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 89
Page 90
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 90
Page 91
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 91
Page 92
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 92
Page 93
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 93
Page 94
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 94
Page 95
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 95
Page 96
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 96
Page 97
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 97
Page 98
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 98
Page 99
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 99
Page 100
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 100
Page 101
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 101
Page 102
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 102
Page 103
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 103
Page 104
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 104
Page 105
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 105
Page 106
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 106
Page 107
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 107
Page 108
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 108
Page 109
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 109
Page 110
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 110
Page 111
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 111
Page 112
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 112
Page 113
Suggested Citation:"Chapter 3 - Findings and Applications." National Academies of Sciences, Engineering, and Medicine. 2021. Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements. Washington, DC: The National Academies Press. doi: 10.17226/25971.
×
Page 113

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

25 Findings and Applications This chapter presents the results from the three phases of this research study. The results of the literature review are presented as they relate to the identification of existing and developing quality tests. The results of the laboratory and field testing are summarized and analyzed with respect to the ability of each test to detect changes in material behavior related to changes in stabilizing/recycling agent content and type, presence of an active filler, and curing time. The results of the ruggedness study and ILS are also presented. 3.1 Current and Emerging Quality Tests The research team sought to identify current and emerging quality assessment and process control tests for cold recycled materials where emulsified asphalt or foamed asphalt serves as the stabilizing/recycling agent. This work was completed by reviewing the available literature, conducting a review of agency specifications, and conducting an online stake- holder survey. 3.1.1 Literature Review The literature survey showed that the quality of recycled materials is most often assessed in the field during or just after construction by measuring the density and moisture content of the recycled layer. Other tests that are commonly performed on materials collected as either loose samples or plant-produced, laboratory-compacted specimens include the gradation of the recycled mixture, the determination of the stabilizing/recycling agent content, and various strength tests such as indirect tensile strength (ITS) and Marshall stability (Chen and Jahren 2007, Asphalt Academy 2009, Wirtgen 2010, Diefenderfer et al. 2015). Properties of the compacted recycled layer, such as stiffness, penetration, and shear resistance, and field versions of current laboratory-based tests have also been reported (VanFrank 2015). The density of the recycled layer is most often measured using a nuclear density gauge. The density achieved during construction is usually compared to either a laboratory- based reference density (determined during mix design) or the density achieved for a test section completed prior to full-scale construction. Other options include comparing the field density to a field-measured standard such as a modified Proctor test, but this is less commonly employed. Density measurements have been shown to be somewhat correlated with stiffness properties of recycled materials (Schwartz et al. 2017), and the experience of the recycling community has suggested that poor density generally leads to poor material quality (Nataatmadja 2001, ARRA 2015). However, recent research has reported that density and permanent deforma- tion resistance in the laboratory may not be well correlated (Bowers et al. 2018). The moisture content of the recycled layer is most often measured using a nuclear density gauge or collecting a sample from the field for analysis in the laboratory by oven drying. The advantage to the former is that the nuclear density gauge is already used at the project for density measurements. There are a host of non-nuclear methods available that are reported less in the literature. These methods include electro- magnetic methods (time-domain reflectometry/dielectric, frequency domain reflectometry/capacitance, amplitude domain reflectometry/impedance, phase transmission, and time-domain transmission) and tensiometric methods (matric suction). These alternatives can be helpful in that moisture measurements using the nuclear density gauge are affected by the hydrogen present in the asphalt binder. Moisture content measurements using the nuclear density gauge are often corrected by using an offset calculated from oven- based drying in the laboratory. Deficient moisture content may inhibit the ability to achieve the compaction necessary for good performance. Excessive moisture not only may reduce the ability to achieve compaction but may also significantly delay the placing of C H A P T E R 3

26 any subsequent layer. Measurements of the recycled material’s moisture content are often cited in specifications for deter- mining the appropriate time to open a recycled layer to traffic or surfacing; however, it is unclear as to what maximum moisture content is permissible that will still achieve satisfac- tory performance of the constructed layer. Previous studies have suggested that as a recycled material loses moisture, the particle bonds are enhanced and strength properties increase (Mohammad et al. 2003, Bemanian et al. 2006, Lee and Im 2008, Lee et al. 2009, Kim et al. 2011, Tebaldi et al. 2014). Therefore, it is a concern of agency practitioners that a recy- cled layer not be surfaced or released to traffic too early so that permanent deformation of the recycled layer does not occur. However, the moisture content is truly only a proxy, as the desired parameter is the sufficiency of the recycled layer to carry loading without damage. Agency specifications often require that a defined moisture content be achieved prior to surfacing or release to traffic. These requirements vary from a 2% reduction in the as-placed moisture content to a moisture content of half the optimum to as low as an in-situ measured value of 1% to 2% (Kim et al. 2011, Texas Department of Transportation [DOT] 2018, Woods et al. 2012). Rather than measuring the moisture content, certain agency specifications may require the con- tractor to wait a predetermined time prior to surfacing or release to traffic. These wait times may range from a few hours to more than 2 weeks (Woods et al. 2012). A problem with such wait time provisions is that they do not address ambient conditions that can significantly affect curing. In addition to nuclear gauge measurements of the recycled material’s moisture content, the oven drying method may be the most direct method. However, the results of the test are not available for up to 24 hours, making it more of a quality assurance than a quality control procedure. Some contractors and agencies may use an on-site propane burner, microwave drying, or calcium carbide gas pressure meter (Speedy) method for drying samples to obtain results in less time. Even with these additional methods, the paving process will be some distance away by the time the moisture content measurements are available, and the opportunity to provide a corrective action may be lost. Lee and Im (2008), Lee et al. (2009), and Kim et al. (2011) explored the relationships between curing time and the results of various strength tests. After showing that recycled material strength generally increased with reductions in moisture content, they evaluated methods for assessing the moisture content to assess field curing. A handheld capaci- tance sensor and a portable time-domain reflectometry unit were used to assess the moisture content, along with an SSG and LWD to assess the stiffness and measure the modulus of the recycled layers, respectively. Measuring the stiffness of the recycled material or the load carrying capacity of a pavement structure with recycled material layers is most often done using a falling weight deflectometer (FWD) (Diefenderfer and Apeagyei 2011). Deflection testing with an FWD is most often performed several days to weeks after construction as the stress applied can cause plastic deformation in a recently completed recy- cled layer (Schwartz and Khosravifar 2013). In addition to deflection testing with an FWD, the literature reports several other devices that could be used to assess the stiffness of the recycled material. These devices include the SSG, the LWD, the portable seismic pavement (or property) analyzer (PSPA), and Clegg hammer (Wilson and Guthrie 2011). In addition, penetration resistance may be assessed using a DCP (Sebesta et al. 2009). The LWD is quickly becoming a popular device for field- based deflection testing of unbound layers primarily because of its portability and because its output is a fundamental engineering property (stiffness). An LWD operates on a prin- ciple similar to that of an FWD but at a much lower stress level. The LWD consists of a falling weight traveling on a guide rod with both attached to the center of a load plate. A deflection sensor within the load plate measures the deflec- tion of the pavement layer caused by the falling weight, and from this the elastic modulus is calculated. Schwartz and Khosravifar (2013) and Meocci et al. (2017) used the LWD to study the placement of CCPR materials. They reported that the LWD was useful for this purpose and had the advan- tage of providing a direct engineering property (i.e., material stiffness). Schwartz and Khosravifar (2013) used the LWD to eval- uate the field stiffness gain of a CCPR material placed in Maryland. They found that the LWD was able to quantify the stiffness gain of the CCPR material, but they encountered operational issues with the LWD used in the study. By means of finite element analysis, the authors found that the zone of influence from the LWD was about twice the diameter of the loading plate. This is an important concept for all measure- ment devices but was seldom reported in the literature that was reviewed. Betti et al. (2017) also used the LWD to assess the stiffness of recycled layers soon after construction and followed up with the FWD to assess the stiffness at later curing stages. Schwartz and Khosravifar (2013) discussed an aspect of deflection testing that is often overlooked: the zone of influ- ence of the measurement. The zone of influence is the area within the pavement structure that responds to the stress applied during the test. The authors point out that the zone of influence from LWD testing may extend well beyond the depth of the recycled layer, especially for thinner applica- tions. Another consideration is the ability of the deflection device to differentiate between the recycled layer and any

27 adjacent asphalt mixture layer. It may be difficult to isolate completely the stiffness properties of the recycled layer in a multilayered pavement structure, as described by Diefenderfer et al. (2012, 2016b). The SSG was developed as a low-cost and portable tool to assess the stiffness of compacted soils. The SSG operates by applying a vibrating force to the soil surface through a seating foot. The force and displacement–time history are measured and used to calculate the stiffness of the soil. Scullion (2002) used the SSG to study the stiffness change in soil–cement bases that were pre-cracked in an effort to reduce reflection cracking. Alshibli et al. (2005) and White et al. (2013) used the SSG as one of the tools to monitor the change in modulus with time and other properties for various compacted layers. Schwartz and Khosravifar (2013) used the SSG to assess the stiffness of a CCPR layer placed in Maryland. By the fourth day after construction, they found the stiffness of the CCPR layer to be greater than the upper limit that can be measured by the SSG. Although reaching a measurement limit could be a concern for long-term monitoring of a recycled section, the time frame fits within the time window expected by the present research team for tests used in this study. Woods et al. (2012) show that an SSG could be used to assess the stiff- ness of a recycled layer such that the contractor could use the measurement to determine when an overlay should be placed. However, their study did not suggest stiffness values or changes in stiffness values that would identify optimal timing of the overlay. The collection of cored samples from a project is perhaps the best way to ensure that the strength properties of only the recycled layer are assessed. Although core sampling permits direct testing of the materials in their as-placed and field- cured condition, there are several significant drawbacks. Core sampling is not a reliable method for obtaining test materials at early ages of the recycled layer. It is likely to be difficult to retrieve the cored sample until the recycled layer has cured sufficiently to withstand the coring process, approximately 4 weeks to 6 weeks after construction (Asphalt Academy 2009). With respect to testing once a core sample is retrieved, the most suitable test conditions are yet to be established. Recent research has shown that confinement has a signifi- cant influence on the stiffness and permanent deformation resistance properties of recycled materials (Diefenderfer and Link 2014, Schwartz et al. 2017), and this can be difficult to replicate for certain test geometries (especially indirect tensile). The ability to collect a sample of adequate size is also a concern and has given rise to several studies looking at alternative specimen sizes (Li and Gibson 2013, Bowers et al. 2015, Schwartz et al. 2017). Another method for assessing the strength of the recycled layer is proof rolling. This procedure is most often used as a process control tool but is occasionally specified for accep- tance. Its use as part of a process control or quality assurance program could be expanded if the test was standardized. Proof rolling can effectively identify deficient material issues and inconsistencies; however, there is no standard method to apply the test, and thus results are not currently transferable from one project to the next. A review of the literature found that work has been performed in Ohio, Pennsylvania, Texas, and Wisconsin and by the Federal Aviation Administration to standardize the process (Texas DOT 2001, City of Columbus 2012, Ohio DOT 2013, Federal Aviation Administration 2014, Pennsylvania DOT 2016). Crovetti and Schabelski (2002) dis- cuss the development of an instrumented tandem-axle dump truck (instrumented to measure the resulting deformation), which, when loaded, can be used to automate a proof roll test. These references document proof rolling using either a heavy roller or a loaded truck with a gross vehicle weight ranging from 30 tons to 36 tons. Unacceptable permanent deforma- tion values were noted as ranging from 0.5 in. to 1.5 in. Ground penetrating radar (GPR) is a well-known but not frequently used device that can be used to assess the thick- ness of pavement layers (Maser and Scullion 1992, Maser et al. 2006, Holzschuher et al. 2007). For recycled pavement projects, GPR is used to assess the pavement thickness either prior to recycling operations to assist with the design of the pavement structure or after recycling in forensic applications (Loizos and Plati 2007, Mallick et al. 2007). Measuring the thickness of the layers postconstruction is especially benefi- cial during the analysis of FWD test results where changes in thickness can influence the calculated pavement structural capacity, as discussed in more detail in Diefenderfer and Apeagyei (2011). In addition to thickness measurements, GPR has been used more recently to assess the density uniformity of asphalt mixtures by correlating the measured dielectric constant to the air-void content from collected cores (Al-Qadi et al. 2010, Leng et al. 2012). To date, this work has been performed on hot-mixed asphalt mixtures and not on cold recycled mixtures, but the principles of operation would be similar. Although not assessing the air-void content per se, GPR could be used as a tool to evaluate the uniformity of the layer by observing the changes in correlated air-void content. Recent advances in this area have been commercialized as a rolling density meter, which includes multiple GPR antennas either mounted on a pushcart or attached to a vehicle. Figure 3.1 shows examples of a GPR system equipped for pavement thickness testing and pavement density testing. The DCP has also shown promise in the evaluation of unbound, fine-grained soils (Chen et al. 2001, Jersey and Edwards 2009, Kazmee et al. 2017) and recycled materials (Alghamdi 2016, Sargand et al. 2016). The DCP consists of a weight attached to a metal rod that ends with a penetra- tion cone. The weight is dropped from a given height onto an

28 anvil that pushes the cone into the test material. A connected scale measures the distance of penetration per drop, and this penetration has been correlated with strength param- eters such as the California bearing ratio (Alghamdi 2016). Although there are many publications citing the use of DCP in unbound materials, its use for bound recycled materials is less common. Tingle and Jersey (1999) and Siddiki et al. (2008) provide guidance on using the DCP for acceptance of compacted material. The PSPA has been used to measure fundamental prop- erties of pavement layers by using wave propagation tech- niques. This is accomplished with a high-frequency wave propagation source and receiver accelerometers. The PSPA measures the seismic modulus of the surface pavement layer. Williams and Nazarian (2007) state that a primary benefit of measuring the moduli of pavement layers using the PSPA is the portability of the device; the main drawback is that the stress state imposed on the tested material is quite dif- ferent (and significantly less) than that applied by a wheel load. Williams and Nazarian (2007) discuss the relationship between seismic modulus (at low strain levels) and resilient modulus (at higher strain levels) and report the relationship at low strain levels. From this, the prospects of using the PSPA for recycled materials (assuming a granular-like behavior) is promising, although it depends on the commercial availability of the device. A procedure for using the PSPA on FDR layers is presented by Mallick et al. (2007). A test to assess the resistance to raveling has been used for some time in a laboratory setting. The specification, ASTM D7196, describes the procedure to assess the raveling resistance of CIR materials using a laboratory stand mixer applying a load to a rubber hose. The rubber hose abrades the surface of the CIR specimen, and mass loss is measured. This test is designed for a laboratory setting, but an alter- native might be developed to make it field ready. These alter- natives could be as simple as using the turning movement caused by the steering axle of a loaded truck to an oscillating/ rotating load applied in a way that is like the laboratory test. Piratheepan et al. (2012) discussed a laboratory testing program to investigate the effect of curing on the raveling resistance of CIR mixtures in accordance with ASTM D7196. They stated that most agency specifications allow a maximum of 2% mass loss. Other tests that were identified but that appear to need further development include a modified version of the labo- ratory abrasion test detailed in ASTM C944 using a studded wheel (Dong et al. 2010), a modified version of a cohesion test detailed in ASTM D3910 (Dong et al. 2010), a wire brush test detailed in ASTM D559, and a modified version of the emulsion-surface–treated sweep test detailed in ASTM D7000 (Johannes et al. 2011). 3.1.1.1 Recent Developments in Field Quality Testing of Recycled Materials A series of research reports published by the Utah DOT details research investigating novel methods for assessing the quality of CIR mixtures (VanFrank et al. 2014, VanFrank 2015). Included among these methods are a shear vane test to assess stability properties, an MH-based field test to assess deformation resistance, and the use of the DCP to study penetration resistance. These reports are discussed herein in a separate section rather than by test type as they are the most comprehensive studies found in the literature with direct application to this study. Based on these studies, the Utah DOT states in its Guidelines for Evaluation, Mix Design, and Field Acceptance of Cold Recycling of Asphalt Pavements Using Solventless Emulsion that cold recycled materials shall be assessed for release to traffic by using the results of the shear vane and Marshall field tests (Utah DOT 2007). Figures 3.2 and 3.3 show a photograph and a schematic of the modified shear vane used by the Utah DOT, respec- Figure 3.1. GPR system for pavement thickness testing (left) and pavement density testing (right).

29 tively. This modified shear vane is much more robust than similar products used for unbound materials and is made from a 3-in.-square steel washer having a 3⁄16 in. thickness, a 5⁄8-in.-diameter bolt that has the end fashioned to a point, and 1⁄8-in.-thick steel plate flanges welded to the bolt and the steel washer. The shear vane is hammered into a CIR layer using a 5-lb hammer until the top washer is flush with the surface. A torque wrench is attached to the bolt head using a standard socket, and torque is applied such that the end of the torque wrench travels 90° in 10 seconds. The greatest torque read on the dial of the torque wrench prior to the material breaking loose is recorded as the shear value (in foot-pounds) along with the pavement temperature at a depth of 2 in. VanFrank (2015) stated that the recycled layer was ready for traffic when a shear value of 30 ft-lb was obtained during field testing. The Utah DOT (2017) also discussed the use of an MH to assess the time to release to traffic and the completion of compaction efforts. The test was conducted using an MH (as specified in AASHTO R 68) in the field, and 50 blows were applied to the surface of the recycled layer. The depth of penetration was measured with respect to the level of the undisturbed surface. The Utah DOT (2017) stated that the layer is not ready for final rolling and opening to traffic if the depression is greater than 10 mm, the height of the lateral deformation is greater than 5 mm, or if bleed water appears. It is not clear if the test should be used only as a release to traffic assessment or as an indication that the layer is ready for final compaction in addition to a release to traffic. VanFrank (2015) also noted that the DCP was used as a third assessment tool for recycled materials. From field studies it was found that when the DCP penetration was less than 10 mm per blow, the recycled layer was ready for release to traffic. It is not yet clear why the agency did not include DCP testing in the instruction manual. However, VanFrank (2015) stated that during field testing, it was found that both the shear vane and DCP criteria could be met while the recycled material was still in a plastic state—that is, still susceptible to flow instability. The author attributed the inability of the shear vane to identify this as because of the localized nature of the testing and the high degree of particle displacement around the shear vane edges. The author found that including the MH field test allowed assessment of the bulk particle movement from a larger stress influence. VanFrank (2015) also noted that multiple tests were needed to achieve the desired result of identifying a time for release to traffic. 3.1.2 Stakeholder Survey The following details the results of the responses to the online stakeholder survey. The survey link was distributed to agencies via the AASHTO Committee on Materials and Pavements and to attendees of regional and national pavement recycling conferences. There were 81 responses to the survey. 3.1.2.1 Demographics The demographics of the survey respondents are shown in Figures 3.4 through 3.6. Figure 3.4 shows that although 23% of the respondents had less than 2 years of experience with pavement recycling techniques, 47% had more than 10 years of experience. Figure 3.5 shows that 74% of the survey respondents identified their organization as a state or local agency, 15% as a member of industry (including con- tractors, suppliers, and testing firms), and 10% as academic. Figure 3.6 shows the geographic location of the survey respon- dents; 44% of the respondents reported their location as either northeast or northcentral United States. Source: Utah DOT 2017. Figure 3.2. Utah DOT shear vane used for cold recycled materials. Source: Utah DOT 2017. Figure 3.3. Schematic of Utah DOT shear vane used for cold recycled materials.

30 3.1.2.2 Stabilizing/Recycling Agent and Active Filler Use Figure 3.7 presents the information gained about the use of particular stabilizing/recycling agents and shows that emulsified asphalt was used by the survey respondents more often than foamed asphalt. Figures 3.8 through 3.10 show that cement was the most prevalent active filler used by the survey respondents regardless of whether the stabilizing/recycling agent was emulsified or foamed asphalt. 3.1.2.3 Turnaround Time Figure 3.11 shows the responses from the respondents when asked about the maximum acceptable turnaround time for a test that determines when a cold recycled pavement may be opened to traffic or surfaced. Figure 3.11 shows that most respondents identified that the maximum allowable turnaround time for a test to determine opening to traffic was less than 4 hours (65%). Most of the respondents iden- tified that the maximum allowable turnaround time for a test to determine the time to surfacing was within 1 day (74%). Based on this finding, the research team focused on testing that could be completed within the first 24 hours following construction of a recycled layer. 3.1.2.4 Challenges in Implementing Specifications In replying to the question about challenges encountered with implementing public-agency specifications for CR, a lack of experience was noted as an area of concern by 43% of the respondents. (Of agency respondents, 54% indicated that agency experience was a concern, and 55% of industry respondents agreed; 27% of industry respondents cited industry experience as a concern, and 56% of agency respon- dents agreed.) Challenges regarding tests and specifications were identified by 39% of the respondents. Additional chal- lenges identified are shown in Figure 3.12. 3.1.2.5 Recommended Tests and Suggested Changes to Existing Tests The survey asked the recipients to indicate any recom- mended tests for process control or acceptance, time to trafficking/surfacing, and long-term performance. From the responses, several suggestions were identified that influ- enced the tests selected for the laboratory testing. These suggestions included density and moisture assessments, deflection testing, penetration resistance, proof rolling, and field shear testing. 20 years or more, 17% 10-20 years, 30% 5-10 years, 14% 2-5 years, 16% Less than 2 years, 23% Figure 3.4. Stakeholder survey respondent experience with recycling techniques. Testing Inspection, 1% Local Agency, 3% Design Engineering, 3% Pavement Recycling Contractor, 3% General Highway Contractor, 4% Materials Supplier, 4% Academic, 10% State Agency, 71% Figure 3.5. Stakeholder survey identified organization. Northwest US, 12% Southwest US, 10% Northcentral US, 24% Southcentral US, 7% Northeast US, 20% Southeast US, 16% Canada, 8% Other, 2% Figure 3.6. Stakeholder survey respondent geographic work location.

31 0% 5% 10% 15% 20% 25% 30% FDR CIR CCPR Pe rc en ta ge o f R es po ns es Recycling Process Foam Emulsion Figure 3.7. Stabilizing/recycling agent by recycling process. 0% 10% 20% 30% 40% 50% 60% 70% Cement Lime Others Pe rc en ta ge o f R es po ns es Chemical Additive/Active Filler Foam Emulsion Figure 3.8. Active fillers used with FDR. 0% 10% 20% 30% 40% 50% 60% 70% Cement Lime Others Pe rc en ta ge o f R es po ns es Chemical Additive/Active Filler Foam Emulsion Figure 3.9. Active fillers used with CIR.

32 0% 10% 20% 30% 40% 50% 60% 70% Cement Lime Others Pe rc en ta ge o f R es po ns es Chemical Additive/Active Filler Foam Emulsion Figure 3.10. Active fillers used with CCPR. 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 0-30 min 30-60 min 1-4 hours 4-12 hours 1 day 2 days 3 days 7 days 14 days Pe rc en ta ge o f R es po ns es Maximum Acceptable Time open to traffic time to surface Figure 3.11. Maximum acceptable turnaround time for a test that determines when a cold recycled pavement may be opened to traffic or surfaced. 3.1.2.6 Evaluation Factors The survey recipients were asked to rate the importance of several broadly characterized evaluation factors to help the research team determine the priority of tests proposed for laboratory testing. The respondents rated the importance of the following evaluation factors across three time frames (initial, short term, and longer term): • The time until results are available for the contractor and agency staff to make decisions; • Test location (on the road or in a laboratory); • Condition of the material (loose/molded or in situ); • Equipment availability, cost, and portability; • Application of test results to mix design, construction quality, and validation of the design intent; • Level of skill required by the test operator; • Stated accuracy, precision, and bias (APB) of the test; and • Applicability of results across CIR, CCPR, and FDR materials. The survey recipients were asked to rate the importance of each evaluation factor as being very important, some- what important, or not important. The three ratings were assigned a numerical value, and the average rating for each evaluation factor was calculated. The ranked order from the

33 survey responses was used in the evaluation of candidate tests. Table 3.1 shows the ranking of each evaluation factor by time frame, where a value of 1 indicates the highest-rated factor and a value of 8 indicates the lowest-rated factor. Given the objectives of this study and the results of the stakeholder survey, the results based on the initial and short-term time frames are most relevant. 3.1.2.7 Preferred Location for Time to Trafficking/Surfacing Test The survey recipients were also asked to identify their preferred location for time to trafficking/surfacing test. Options available were field, laboratory, and no preference. Logically, the field location was the option with the highest percentage of responses at 73%. The laboratory location and no preference of location were selected by 7% and 20% of the respondents, respectively. This finding helped to direct the study toward particular tests for Phase II. 3.1.2.8 Tests Most Often Used for Suggested Properties Survey recipients were asked to identify those tests that were most often used for determining the deformation resistance, raveling resistance, density, stiffness, and curing initiation; the responses are shown in Figures 3.13a through 3.13e, respectively. Figure 3.13 shows that most respondents did Previous unsuccessful experiences, 9% Lack of quality tests with quick results, 14% Unreasonable quality tests requested, 2% Ability to meet the required quality test requirements, 5% Lack of specification uniformity across different agencies, 9% Constraints in the means and methods, 3% Excessive time required before opening to traffic/surfacing, 9% Lack of experienced local contractors, 21% Lack of agency experience, 22% Other, 5% Figure 3.12. Challenges encountered with implementing agency specifications for pavement recycling. Evaluation Factor Ranking Initial Short Term Longer Term The time before results are available for the contractor and agency staff to make decisions with 2 1 3 Test location (on the road or in a laboratory) 5 2 8 Condition of the material (loose/molded or in situ) 6 3 7 Equipment availability, cost, and portability 3 4 3 Application of test results to mix design, construction quality, and validation of the design intent 1 5 1 Level of skill required by the test operator 7 6 3 Stated APB of the test 3 7 2 Applicability of results across CIR, CCPR, and FDR materials 8 8 6 Note: A lesser number indicates a higher-rated factor. Table 3.1. Survey ranking of evaluation factors by time frame.

34 (a) Deformation Resistance (b) Raveling Resistance (c) Density (d) Stiffness (e) Curing Initiation Figure 3.13. Tests most often used to determine (a) deformation resistance, (b) raveling resistance, (c) density, (d) stiffness, (e) curing initiation.

35 not have a particular test for determining the deformation resistance, raveling resistance, stiffness, or curing initiation. Density was most often assessed by the nuclear density gauge. For those properties where no particular test was mentioned by most of the respondents, one or two particular tests were mentioned more than others. DCP was most often selected for deformation resistance, and the raveling test using a stand mixer was most often selected for raveling resistance. FWD was most often selected for stiffness. When identify- ing a test for curing initiation, the respondents most often selected “none”; “other” was selected the next most often. For each property assessed, the respondents may have chosen “other” to indicate a test that was not on the list provided by the research team. The survey recipients were asked to identify the “other” test methods, and the results are shown in Table 3.2. 3.1.3 Specification Review A total of 83 Canadian/U.S. provincial/state and local agency specifications were reviewed, as indicated in Fig- ure 3.14. These specifications included only those for asphalt- based stabilizing/recycling agents. Figure 3.14 also shows the geographical distribution of specifications by recycling process; 24 FDR, 45 CIR, and 14 CCPR specifications were reviewed. The reviewed specifications generally contained the following major sections: description of the process, materials, equipment, methods, quality measures, weather, curing, and measurement and payment. The degree of detail in each specification varied widely, from a short construc- tion section and measurement and payment to more detail including all sections listed previously and sections on a preconstruction meeting, a preconstruction quality control plan, a detailed description of equipment requirements, and acceptance testing. This is similar to findings by Stroup- Gardiner (2011) and Salomon and Newcomb (2000). The following discussion focuses on summarizing the post- construction quality characteristics and measures found in the reviewed specifications. The methods used to assess material quality after construction varied widely by agency but most often included a measure of density, moisture content, curing time, strength value, and gradation. Figure 3.15 shows the measures most often required for quality of the constructed layer. Density was the most popular measure (included in 94% of the specifications reviewed), followed by gradation, moisture content, and curing time. (Gradation will not be discussed further in this section since it is a laboratory-measured property, and the results of the stakeholder survey indicated a clear preference for field-based tests.) Figures 3.16 through 3.18 show the distribution of construction quality characteristics for FDR, CIR, and CCPR respectively. As reflected in the cumulative analysis shown in Figure 3.15, density is the most commonly prescribed quality characteristic for all three recycling processes. Given the relatively lower percentages for performance tests (e.g., DCP, stability, test/proof rolling), few specifications apply performance testing to field acceptance of the recycled layer at this time. Specifications requiring moisture content readings as an indicator of curing or trafficking/surfacing readiness often required the moisture content to be in a range of from 1% to 3.5% or to be a percentage of the optimum moisture content established during mix design. Of all reviewed specifications (29 specifications), 35% called for a moisture reading of 2% or less before a recycled layer could be surfaced. Of the specifications referencing the optimum moisture content from design, 15.7% (13 specifications) required a moisture reading of less than 50% of the optimum moisture content, and some specifications, which were placed under the “other” category, required a moisture reading to be within some range of the optimum moisture content. Table 3.3 shows the required moisture content and the percent of specifications that required a specific moisture content as a percent of all specifications that required moisture readings. For specifi- cations where an either/or scenario about moisture content was stated (e.g., either percent of total moisture content or 50% of target optimum moisture content), both cases were counted; thus, the percent occurrences in Table 3.3 add to more than 100%. Based on the proportion of agency recycling specifications that included it as a characteristic, estimating that a newly placed recycled layer is cured is a concern. However, most definitions of “curing” relate to a moisture content measure- ment rather than to curing itself being a true measure of the Property Recommended Test Deformation resistance Various lab-based tests including IDT, triaxial, Hamburg Observation of the rolling process (visual observation) Raveling resistance ASTM D7196 (mostly used during mix design only) Density – Stiffness DCP Nuclear density gauge Curing initiation CoreLokMoisture content by field drying (AASHTO T 255) Table 3.2. Tests recommended in stakeholder survey.

36 (a) (b) (a) (b) Figure 3.14. Reviewed agency recycling specifications, shown by shaded area: (a) all recycling processes, (b) FDR, (c) CIR, (d) CCPR.

37 22% 2% 94% 60% 64% 13% 6% 16% 67% 0% 20% 40% 60% 80% 100% Pe rc en t o f S pe ci fic at io ns w ith Te st R eq ui re m en t ITS DCP Density Curing Time Moisture Content Marshall Stability Raveling Stability Test/Proof Rolling Gradation Test Type Figure 3.15. Distribution of constructed quality characteristic including all recycling processes. 29% 4% 92% 46% 63% 0% 0% 46% 71% 0% 20% 40% 60% 80% 100% Pe rc en t o f S pe ci fic at io ns w ith Te st R eq ui re m en t ITS DCP Density Curing Time Moisture Content Marshall Stability Raveling Stability Test/Proof Rolling Gradation Test Type Figure 3.16. Distribution of constructed quality characteristic for FDR. 20% 2% 96% 64% 69% 18% 7% 4% 67% 0% 20% 40% 60% 80% 100% Pe rc en t o f S pe ci fic at io ns w ith Te st R eq ui re m en t ITS DCP Density Curing Time Moisture Content Marshall Stability Raveling Stability Test/Proof Rolling Gradation Test Type Figure 3.17. Distribution of constructed quality characteristic for CIR.

38 14% 0% 93% 71% 50% 21% 14% 0% 64% 0% 20% 40% 60% 80% 100% Pe rc en t o f S pe ci fic at io ns w ith Te st R eq ui re m en t ITS DCP Density Curing Time Moisture Content Marshall Stability Raveling Stability Test/Proof Rolling Gradation Test Type Figure 3.18. Distribution of constructed quality characteristic for CCPR. Moisture Content, % Percent of Occurrences ≤1 2.4 ≤1.5 6.0 ≤2 26.5 ≤2.5 6.0 ≤3 6.0 <3.5 1.2 50% of OMC 15.7 Other 6.0 Not specified 36.1 Note: OMC = optimum moisture content. Table 3.3. Required moisture content before resurfacing and percent of occurrences. Curing Time Percent of Occurrences 1 hour 1.2 2 hours 3.6 4 hours 1.2 6 hours 1.2 2 days 8.4 3 days 21.7 4–7 days 16.9 >7 days 12.0 Table 3.4. Required curing times and percent of occurrences. ability to withstand loading without damage (by premature rutting or raveling) as a function of moisture content. A speci- fied curing time prior to surfacing was required by 60% (50 specifications) of the reviewed agency specifications, and 64% (53 specifications) required a specific moisture content or moisture reduction. Thirty-six percent (30 specifications) of reviewed agency specifications stated that both a cure time and moisture content requirement must be satisfied prior to surfacing of the recycled layer. Of the specifications that listed a required curing time, it ranged from 1 hour to multiple days, with a maximum of 14 days. Table 3.4 provides various curing time durations and subsequent percent of occurrences for the reviewed specifications. Assuming a curing time of zero for those specifications that did not specify a required curing time, 52% (43 specifications) of the reviewed specifications required a cure time of between 2 to 14 days. In Table 3.4, there are cases where a specifica- tion called for curing times that bridged multiple represented curing periods (e.g., a specification called for a minimum of 2 days of curing and a maximum of 7 days). In these cases, the specification would be counted in each of the representative curing time categories (e.g., categories of 2 days, 3 days, and 4 to 7 days for a required 2 to 7 days of curing). This causes the total percentage in the Table 3.4 “Percent of Occurrences” column to be greater than 100%. 3.2 Candidate Tests After the literature review and the stakeholder survey, the research team worked to develop a list of potential (or candidate) tests. This was started by listing those material properties thought to describe best the condition of a recycled layer at initial, short-term, and longer-term time frames. Next, a list of potential (or candidate) tests that could describe these properties was created. The two lists were grouped based on the tests’ ability to assess properties of the recycled layer at three time frames: initial, short term, and longer term. The three time frames were not rigidly defined but instead were thought of as being soon after compaction, within the first 1 to 2 days, and longer than 2 days, respectively. The three time frames are described in more detail as follows: • Initial tests: most often used for process or construction quality control purposes during and immediately after

39 compaction of the recycled layer. Examples include in-situ moisture content, active filler content, recycling depth, grading, compaction density, and initial stiffness gain. • Short-term tests: most often used to quantify the progres- sion of curing and can indicate the relative performance of the surface of the recycled layer. Examples include stiff- ness gain, penetration/deformation resistance, raveling resistance, and moisture content. • Longer-term tests: used to determine if the engineering properties of the recycled layer meet the design intent and to obtain an indication of long-term performance. The results of longer-term tests have been used to develop and refine mechanistic–empirical design procedures for in-place recycled pavements. An example is material stiff- ness characterization. The research team initially listed as many relevant candi- date tests (and the properties they describe) as possible that could be used to assess not only time to opening or surfacing but also material quality. The research team intentionally did not include any test that used visual observation as the basis of measurement; only tests with quantifiable results were included. This list of material properties and their associ- ated candidate tests for initial, short-term, and longer-term properties is shown and described in Tables 3.5 through 3.7, respectively. Some material properties and tests appear across multiple time frames. The research team identified candidate tests with the knowledge that some tests would be included only for completeness, whereas others would progress on to evaluation in the laboratory testing. 3.3 Selected Tests The material property/candidate test lists were further refined by evaluating each material property in terms of several evaluation factors, including the following: • The time before results are available to the contractor or agency; • Test location (laboratory or in situ); • Condition of the material (loose, molded, or in situ); • Equipment availability, cost, and portability; • Application of test results to mix design, construction quality, and validation of the design intent; • Skill level required by the test operator; • Stated APB of the test; and • Applicability of the test for CIR, CCPR, and FDR materials. These factors were incorporated into an evaluation matrix that the research team used to objectively evaluate the merit in assessing each material property. The material property was evaluated rather than a particular test because this improved the objectivity of the evaluation and reduced any potential bias toward or against a particular test. The evalu- ation also used a weighting factor (defined as the ranked priority of the evaluation terms from the stakeholder survey; a lower value indicates a higher priority) and a usage level (determined by the research team based on their experience; a lower value indicates a higher usage level). Summing the product of the weighting factor and the usage level from each evaluation term resulted in a total score for each material property. (A lower score indicates a higher justification to include in Phase II.) The evaluation matrices for initial, short- term, and longer-term parameters are shown in Tables 3.8, 3.9, and 3.10, respectively. The ranked results of the evalua- tion matrices assessment are shown for each time frame in Table 3.11. The research team elected to focus on those properties that were most highly ranked (having the lowest score shown in Table 3.11). A cut-off value of 50 was selected, and those properties that had a score under 50 were used to determine the tests that would be evaluated as part of the laboratory testing: density/compaction; stiffness; penetration, deforma- tion, shear resistance/bearing tests; raveling resistance; and in-situ moisture. Specific tests to assess these properties in the laboratory testing included the tests shown in Table 2.2. 3.4 Laboratory Testing To assess the desired properties shown in Table 2.2, a laboratory experiment was developed using compacted slab specimens manufactured from field-produced materials sampled from actual recycling projects. Replicate test slabs from each field project were fabricated and tested in three sets. Each test slab had dimensions of 500 mm in length × 400 mm in width and a thickness of approximately 110 mm. Using multiple slab sets was necessary given (1) the physical size of the tested area from each test, (2) the desire to conduct testing on undisturbed sections of the slab where possible, and (3) the necessity to account for replication. Table 3.12 shows the tests and curing times assessed for each set of slabs. The results of each test are presented in the following sections. The data were analyzed with respect to the ability of the test to provide (1) a low variability among repetitive measurements for a given mixture, (2) a low variability among mixture replicates, and (3) a wide range or spread with respect to curing time and presence of cement. 3.4.1 Density/Compaction The density of each slab was to be assessed using a thin- lift nuclear density gauge. During exploratory testing, the research team compared nuclear density gauge readings with the bulk slab density obtained by dividing the slab mass by

40 Property Assessed Candidate Test(s) Need/Key Measures Typical/Current Practice Existing Standard? Cost to Conduct Test? In-situ moisture content Gravimetric moisture content Nuclear gauge–based moisture content Electromagnetic moisture probe GPR Confirm proper mixing/compaction moisture content Gravimetric test prior to start of project ASTM D6780/ASTM D7830/ASTM D6836 AASHTO T 310/ ASTM D6938 ASTM D7830 No, but research papers exist Low Active filler content Tarp or pan test Confirm correct amount of active filler has been applied Tarp or pan test No, but common practice Low Recycling depth Probe Slit trench Determine that design requirements are met Probe, slit trench No, but common practice No, but common practice Low Gradation Sieve analysis Check that appropriate grading curve is achieved and that there are no oversize particles Sieve analysis on mixture without recycling/ stabilizing agent ASTM C136/AASHTO T 27 Low Curing time Cohesion tester Typically used for measuring curing time in slurry mixtures. Can be used to assess the time for opening to traffic ASTM D3910 with modification; can be done on laboratory- compacted samples or on compacted mat in the field ASTM D3910; can be easily modified and adopted for laboratory and field applications Low Density/ compaction Using nuclear gauge Sand cone Rubber balloon Proxy test for assessing material quality; some correlation with stiffness Nuclear density (ASTM D2950) or similar state- specific specifications ASTM D2950 ASTM D1556 ASTM D216 Low Stiffness Soil stiffness gauge LWD PSPA Clegg hammer c Rapid compaction ontrol device Can be used to determine degree of curing These stiffness tests have been used primarily on research projects for recycled materials. ASTM D6758 (withdrawn in 2017) ASTM E2583/ASTM E2835/draft specs from TPF-5(285) and NCHRP Project 10-84 No, but research papers exist ASTM D5874 No, but agency procedures exist Medium Penetration, deformation, resistance, bearing tests Ball penetration Marshall hammer Dynamic cone penetrometer Vane shear tests Proof rolling Rapid compaction control device Indicator to determine if curing process has initiated and road can be opened to traffic or is ready for paving Variations of these tests have been used for recycled and other materials. ASTM C360 modified ASTM D1559/AASHTO T 245 modified ASTM D6951 ASTM D2573/AASHTO T 223/ASTM D4648 Yes, agency specs No, but agency procedures exist Medium Table 3.5. Assessed properties and candidate tests for initial parameters.

41 Property Assessed Candidate Test(s) Need/Key Measures Typical/Current Practice Existing Standard? Cost to Conduct Test? In-situ moisture content Gravimetric moisture content Nuclear gauge–based moisture content Electromagnetic moisture probe GPR Confirm proper mixing/ compaction moisture content Gravimetric test prior to start of project ASTM D6780/ASTM D7830/ASTM D6836 /ASTM AASHTO T 310 D6938 ASTM D7830 No, but research papers exist Low Stiffness Soil stiffness gauge LWD PSPA Clegg hammer Can be used to determine degree of curing These stiffness tests have been used primarily on research projects for recycled materials. Upper limits may be quickly exceeded. ASTM D6758 (withdrawn in 2017) ASTM E2583/ASTM E2835/draft specs from TPF-5(285) and NCHRP Project 10-84 No, but research papers exist ASTM D5874 Medium Penetration, deformation, resistance, bearing tests Ball penetration Marshall hammer Dynamic cone penetrometer Vane shear tests Indicator to determine if curing process has initiated and road can be opened to traffic or is ready for paving Variations of these tests have been used for recycled and other materials. ASTM C360 modified ASTM D1559/AASHTO T 245 modified ASTM D6951 ASTM D2573/AASHTO T 223/ASTM D4648 Medium Proof rolling Rapid compaction control device Yes, agency specs No, but agency procedures exist Curing time Cohesion tester Typically used for measuring curing time in slurry mixtures. Could be modified for field use? ASTM D3910 with modification. Modified for field use? ASTM D3910 with modification; modified for field use? Low Raveling resistance Stand mixer raveling test Indicator to determine if compacted surface will ravel under traffic ASTM D7196. Can be done in laboratory on prepared/cored specimen or on the road in field. Based on a Hobart mixer. Can be evaluated after the curing time determined by the cohesion tester. Yes, ASTM D7196, see also ASTM C779, ASTM D4060, ASTM C1138, ASTM C1803 Low Material strength Indirect tensile strength Unconfined compressive strength Triaxial resilient modulus Dynamic modulus Check that design intents are being achieved Indirect tensile strength ASTM D6931/AASHTO T 283 ASTM D2166/ASTM D5102/AASHTO T 208 ASTM D4767/AASHTO T 296/T 297 ASTM D3497 (withdrawn in 2009)/AASHTO T 378 Medium/ high Table 3.6. Assessed properties and candidate tests for short-term parameters.

42 Property Assessed Candidate Test(s) Need/Key Measures Typical/Current Practice Existing Standard? Cost to Conduct Test? Stiffness FWD Plate load test Determine expected life by monitoring changes in stiffness over time; relate measured values to intended design assumptions Dynamic response Static response ASTM D4694 ASTM D1194/ AASHTO T 221 High Table 3.7. Assessed properties and candidate tests for longer-term parameters. the slab volume. It was discovered that the nuclear density gauge reported unreasonable density values likely because of the testing of a small slab relative to the size of the nuclear density gauge. Following the exploratory testing, the calcu- lated bulk density was used as the density value for each com- pacted slab. The slab densities are detailed in tables for each test in the following sections. 3.4.2 Moisture Moisture content measurements using a Troxler Model 6760 Moisture Probe were collected on fabricated test slabs at 2 and 72 hours of curing. Details of the mixture proportions for the tested slabs are provided in Table 3.13. The mois- ture readings were given as a frequency rather than as direct moisture content, and the moisture content was determined by oven drying of a sample removed from the test slab. Since this process was destructive, moisture content measurements were collected only at 72 hours of curing once all other tests were completed. Figure 3.19 shows an example of exploratory testing where the moisture probe was used to assess the moisture content of loose RAP material. The RAP was mixed in a bucket mixer with different amounts of water over a moisture content range of approximately 0% to 7.5%. The results shown in Figure 3.19 suggest that the moisture probe could be quite accurate once a calibration process was completed as recommended by the manufacturer. This step is likely most practical for a large project where many moisture readings might be collected. The calibration process for the explor- atory testing was completed within approximately half a day. Also, during the exploratory testing, it was determined that the moisture device could not be driven into the test slab. The manufacturer included a pin and hammer like those used for a nuclear density gauge test conducted in direct trans- mission mode. During the exploratory testing it was found that the hole remaining from DCP testing could also be used to accommodate the moisture device probe so long as the DCP was extracted carefully after testing. Figure 3.20 shows the mean frequency values obtained from the electromagnetic moisture device for all mixtures at 2 and 72 hours. The moisture device probe was inserted into the hole in the test slab following DCP testing. Three replicate measurements were collected by rotating the moisture device approximately 120° about the hole. Figure 3.20 shows that, for most mixtures, there is a separation between the mean frequency values at the two curing periods. In general, the trend from 2 to 72 hours is in the same direction for all mix- tures except Mixture 3. Figure 3.21 shows the variability of the replicate measurements in terms of the coefficient of variation (COV) for all tests at 2 and 72 hours. Figure 3.21 shows a very low COV, with all values less than 6%. Figure 3.22 shows the variability of measurements on replicate slabs, with all values less than a COV of 8%. Figure 3.23 shows the relationship between the moisture probe frequency and the moisture content obtained by oven drying of a sample. Figure 3.23 shows that the correlation between the two is poor. When the moisture probe was loaned to the research team, the manufacturer stated that the probe needed to be calibrated for each mixture. This calibration was not completed for each mixture and would likely have improved the correlation, as demonstrated in Figure 3.19. 3.4.3 Stiffness Stiffness testing at 2 and 72 hours of curing was completed for 16 different mixtures that were fabricated from 12 different source projects. The number of mixtures is greater than the number of source projects because the research team created some additional mixture types by modifying the mixture design from certain source projects. This was done to facili- tate testing of additional mixture types. Two or three replicate slabs were fabricated from most source projects. In a limited number of cases, a single slab was tested because of the amount of material available. For certain source projects, the replicate slabs had different densities. Rather than dis- carding the slabs, they were included to evaluate the ability of the test to capture changes in density. As a consequence, the stiffness testing was performed on 30 test slabs from the 16 mixtures. Stiffness testing was conducting using an SSG and an LWD.

1 2 3 Time to available test results 2 quick medium long 3 1 1 2 2 2 1 Location of test 5 field on-site lab lab 1 1 1 3 1 1 1 Material condition (loose, molded, or in-place) 6 all in situ only loose/molded 2 3 3 3 1 2 1 Equipment required (availability, portability, and cost) 3 3 positive 2 positive 1 positive 1 1 1 1 2 1 1 Application of results (mix design, construction quality, design validation) 1 3 positive 2 positive 1 positive 2 2 3 3 2 1 2 Operator/data skill analysis level required 7 low medium high 1 1 1 1 1 2 2 Accuracy, precision, and bias of the test 3 spec plus APB statement spec but no APB statement no spec 1 3 3 1 1 1 1 Applicability to different materials (CIR, CCPR, FDR) 8 3 positive 2 positive 1 positive 1 1 1 1 1 1 1 46 54 55 61 41 50 43 Penetration, Deformation, Shear Resistance/ Bearing Tests Usage Level for Material Property (Assigned by the Project Team) Stiffness score = sum of weighting factor rank multiplied by usage level Usage Level Evaluation Term Weighting Factor (Ranking from Survey) In-situ Moisture Active Filler Content Recycling Depth Gradation Density/ Compaction Table 3.8. Evaluation matrix for initial properties.

1 2 3 Time to available test results 1 quick medium long 3 2 1 1 3 Location of test 2 field on-site lab lab 1 1 1 1 2 Material condition (loose, molded, or in-place) 3 all in situ only loose/molded 2 2 1 2 3 Equipment required (availability, portability, and cost) 4 3 positive 2 positive 1 positive 1 1 1 1 3 Application of results (mix design, construction quality, design validation) 5 3 positive 2 positive 1 positive 2 1 2 2 1 Operator/data skill analysis level required 6 low medium high 1 2 2 1 2 Accuracy, precision, and bias of the test 7 spec plus APB statement spec but no APB statement no spec 1 1 1 1 1 Applicability to different materials (CIR, CCPR, FDR) 8 3 positive 2 positive 1 positive 1 1 1 1 1 46 46 47 44 60 Evaluation Term Weighting Factor (ranking from survey) Usage Level Usage Level for Material Property (Assigned by the Project Team) score = sum of weighting factor rank multiplied by usage level In-situ Moisture Stiffness Penetration, Deformation, Shear Resistance/ Bearing Tests Raveling Resistance Material "Strength" Table 3.9. Evaluation matrix for short-term properties.

45 Usage Level for Material Property (Assigned by the Project Team) 1 2 3 Time to available test results 3 quick medium long 2 Location of test 8 field on-site lab lab 1 Material condition (loose, molded, or in-place) 7 all in situ only loose/molded 2 Equipment required (availability, portability, and cost) 3 3 positive 2 positive 1 positive 2 Application of results (mix design, construction quality, design validation) 1 3 positive 2 positive 1 positive 1 Operator/data skill analysis level required 3 2 low medium high 2 Accuracy, precision, and bias of the test spec plus APB statement spec but no APB statement no spec 1 Applicability to different materials (CIR, CCPR, FDR) 6 3 positive 2 positive 1 positive 1 49 score = sum of weighting factor rank multiplied by usage level Evaluation Term Weighting Factor (Ranking from Survey) Usage Level Stiffness Table 3.10. Evaluation matrix for longer-term properties. Parameter Time Frame Property Score Initial Density/compaction Penetration, deformation, shear resistance/bearing tests In-situ moisture Stiffness Active filler content Recycling depth Gradation 41 43 46 50 54 55 61 Short term Raveling resistance In-situ moisture Stiffness Penetration, deformation, shear resistance/bearing tests Material strength 44 46 46 47 60 Longer term Stiffness 49 Table 3.11. Results of evaluation matrix ranked by score.

46 Test Curing Time (hours)Slab Set 1 Slab Set 2 Slab Set 3 Moisture 2, 72 Soil stiffness gauge 2, 72 Lightweight deflectometer 2, 72 1, 3, 6, 24 Dynamic cone penetrometer 2, 72 1, 3, 6, 24 Marshall hammer Long-pin shear test 1, 3, 6, 24 Short-pin raveling test 1, 3, 6, 24 Table 3.12. Test, curing time, and slab set information. Mix ID Stabilizing/Recycling Agent Active Filler Process State Agent Content, % Active Filler Content, % Actual Density, pcf No. of Replicates 1 Emulsified asphalt Cement CCPR IN 2.5 1.0 119.6 22 VA 2.5 1.0 131.5 3 3 FDR TX 4.5 1.1 122.9 14 CA 2.5 1.0 127.8 0 5 No cement CCPR NY 3.0 0.0 124.8 130.2 2 (2 densities) 6 VA 2.5 0.0 128 2 7 CIR ON 1.2 0.0 120.5 2 8 FDR IN 2.5 0.0 29 CA 2.5 0.0 127.8 0 10 Foamed asphalt Cement CCPR VA 2.5 1.0 127.6 0 11 CIR CA 2.0 1.0 120.4 212 MA 2.5 1.0 119.4 2 13 FDR TX 2.4 1.5 125.9 214 CA 2.5 1.0 126.6 3 15 No cement CCPR VA 2.5 0.0 127.6 0 16 CIR MI 2.2 0.0 129.8 2 17 WI 2.0 0.0 121.3118.6 2 (2 densities) 18 FDR CA 2.5 0.0 127.8 3 118 Table 3.13. Specimen and mixture details for moisture content testing. y = 96.42x + 5,988.54 R² = 0.97 5800 6000 6200 6400 6600 6800 7000 0 1 2 3 4 5 6 7 8 Tr ox le r Fr eq ue nc y, H z Moisture Content by Oven Method, % Note: Error bars represent plus/minus one standard deviation. Figure 3.19. Measured moisture content (oven method) versus electromagnetic moisture probe results from loose RAP material. 3.4.3.1 Soil Stiffness Gauge Figure 3.24 shows the average stiffness of the mixtures measured using the SSG at 2 and 72 hours of curing. (For clarity, error bars are not shown.) The average value is made up of three tests per replicate, with the number of replicates shown in Table 3.13. As seen in Figure 3.24, the SSG was generally able to capture the effect of curing time. Of 16 mixtures, three mixtures (Mixtures 8, 13, and 14) showed a decrease in stiffness with respect to curing time. In addition, when individual specimens were considered, 10 of 30 specimens had a lower stiffness with respect to

47 6000 6500 7000 7500 8000 8500 0 2 4 6 8 10 12 14 16 18 20 Fr eq ue nc y, H z Mixture ID 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.20. Electromagnetic moisture device results for all mixtures. 0 2 4 6 8 10 C oe ff ic ie nt o f V ar ia tio n, % 0 2 4 6 8 10 12 14 16 18 20 Mixture ID Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.21. Within-test slab variability for electromagnetic moisture device based on replicate measurements. 0 2 4 6 8 10 0 2 4 6 8 10 12 14 16 18 20 C oe ff ic ie nt o f V ar ia tio n, % Mixture ID 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.22. Between-test slab variability for electromagnetic moisture device based on replicate specimens.

48 increased curing time. Although such a trend was unexpected, several possible reasons might have contributed to this outcome. These include the variability of the material and the fact that the high-frequency test range (from 100 Hz to 196 Hz), small magnitudes of applied force (about 9 N), and small applied strains (approximately 0.00005 in./in.) are thought to reduce the ability to couple the material with the test device. Also, the zone of influence of the device is expected to be greater than the thickness of the test slabs. Table 3.14 shows the descriptive statistics of SSG stiffness with respect to curing time. From Table 3.14, the mean SSG stiffness increased with respect to curing time, as expected, showing the ability of the SSG test to capture the effect of curing. The spread of the SSG-measured stiffness was evalu- ated using the interquartile range (IQR). The IQR was con- sidered, rather than the standard deviation, since the IQR is resistant to effects of outliers. Table 3.14 also shows that the SSG stiffness IQR decreased slightly with respect to curing time. This could be expected since other research (e.g., Schwartz et al. 2017) has shown the similarity in stiffness of different mixture/additive combinations at later ages. Table 3.15 shows the descriptive statistics of SSG stiffness with respect to recycling agent and presence of cement as an active filler. The mean SSG stiffness increased when cement was included for both emulsified asphalt and foamed asphalt mixtures, as expected. Table 3.15 also shows that the IQR increased with the presence of cement as an active filler. This suggests that the presence of cement as an active filler could y = 0.0009x - 3.1776 R² = 0.1849 1.5 2.0 2.5 3.0 3.5 4.0 6200 6400 6600 6800 7000 7200 7400 M oi st ur e C on te nt b y O ve n D ry in g, % Moisture Probe Frequency, Hz Figure 3.23. Moisture probe frequency versus measured moisture content (by oven drying) on compacted test slabs at 72 hours of curing. 0 10 20 30 40 50 0 2 4 6 8 10 12 14 16 18 20 SS G S tif fn es s, M N /m Mixture ID 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.24. Stiffness of the mixtures as measured by soil stiffness gauge.

49 have a larger influence on the SSG-measured stiffness of certain mixtures. Further, when the mean SSG stiffness values were compared, those mixtures that included foamed asphalt (both with and without cement as an active filler) tended to be stiffer than those mixtures that included emul- sified asphalt; the difference was greater for those mixtures that included cement. The within-specimen variability of the SSG stiffness was evaluated in terms of a COV. As shown in Figure 3.25, the COV for replicate measurements was generally less than 10% (for 54 of 60 conditions; 30 test specimens at two different curing times). The average COV was 4% and 6% for the 2- and 72-hour tests, respectively. Replicate measurements at each curing time were possible only for the SSG and LWD tests. The variability of the SSG stiffness among mixture repli- cates was also assessed via the COV. Figure 3.26 presents the range of COV among the evaluated mixtures. The missing points in Figure 3.26 are due to not having a replicate speci- men for a given mixture. In general, the COV among mixtures was less than 30% (30 of 32 conditions). The high variability seen for Mixture 13 was also observed in other tests consid- ered in this study, as shown in the following sections. The average COV was 14.3% and 18.2% for the 2- and 72-hour testing, respectively. The higher COV at 72 hours was due, in part, to the high variability observed for Mixture 13. A way to evaluate the discrimination potential of the SSG test with respect to curing time is to assess the curing ratio. The curing ratio is defined here as the ratio of the stiffness at 72 hours to the stiffness at 2 hours. Figure 3.27 shows the curing ratio of the mixtures. The SSG measurements indi- cated that a total of three mixtures had a lower stiffness with an increase in curing time (i.e., a curing ratio of less than 1). The curing ratio ranged from 0.69 to 5.36, with an average curing ratio of 1.44. Curing Time Mean, MN/m Minimum, MN/m Quartile 1, MN/m Quartile 3, MN/m Maximum, MN/m Range, MN/m Interquartile Range, MN/m 2 hours 19.9 6.3 14.8 23.1 32.2 25.9 8.3 72 hours 24.2 12.3 20.9 28.3 33.5 21.2 7.4 Table 3.14. Descriptive statistics of SSG stiffness by curing time. Material Combination Mean, MN/m Minimum, MN/m Quartile 1, MN/m Quartile 3, MN/m Maximum, MN/m Range, MN/m Interquartile Range, MN/m Emulsion, cement 21.2 6.3 12.5 27.7 33.5 27.3 15.3 Emulsion, no cement 20.0 11.5 12.8 23.2 30.7 19.2 10.4 Foam, cement 26.3 19.6 22.9 30.2 32.2 12.6 7.3 Foam, no cement 21.0 15.7 17.9 23.2 25.7 10.1 5.3 Table 3.15. Descriptive statistics of SSG stiffness by recycling agent and active filler type. 0 5 10 15 20 25 0 4 8 12 16 20 24 28 32 C oe ff ic ie nt o f V ar ia tio n, % Specimens 2-hour 72-hour Figure 3.25. Within-specimen soil stiffness gauge variability in terms of a coefficient of variation.

50 The generated data were statistically analyzed to investi- gate the effect of the recycled mixture parameters consid- ered in this study on SSG stiffness. An analysis of covariance (ANCOVA) at a confidence level of 95% was used to test for significant factors on the SSG stiffness response among various mixture parameters. The mixture parameters used as factors were process type, recycling agent type and rate, active filler (cement) rate, and curing time. The experiment was a nested design as it was not intended to have a factorial design for the levels of the various factors. In other words, although the recycling agent type (emulsion or foam) was not nested as a factor in the process type (CR or FDR), the recycling agent rate and cement rate were nested as a factor in the recycling agent type. The density was input as a covariate during the analysis. Table 3.16 presents the ANCOVA statistics for the SSG stiffness. It is shown that the SSG stiffness was significantly varied (the p-value was less than 0.05) as a function of curing time and recycling agent content with different recycling agent types. The cement rate could not be estimated and was removed from the analysis by the statistical software (Minitab), potentially because of the interaction effect between the recycling agent rate and cement rate in addition to how the analysis design was set up (i.e., both recycling agent rate and cement rate were nested under recycling agent type). This observation does not suggest that the cement rate was not a significant factor; rather, having recycling agent rate as a statistically significant factor and the interaction effect with cement suggest that cement rate might be a significant factor. Although the data were not checked at a mixture level, it is 0 10 20 30 40 50 60 70 80 90 0 2 4 6 8 10 12 14 16 18 20 C oe ff ic ie nt o f V ar ia tio n, % Mixture ID 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.26. Soil stiffness gauge variability among mixture replicates in terms of a coefficient of variation. 0 2 4 6 8 10 0 2 4 6 8 10 12 14 16 18 20 C ur in g R at io Mixture ID Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.27. Curing ratio of mixtures from soil stiffness gauge.

51 anticipated that the curing time is a statistically significant factor only for the mixtures with cement because the variation in the response values with respect to time for the mixtures without cement did not vary significantly from each other. 3.4.3.2 Lightweight Deflectometer The LWD was used to assess the modulus of each test slab immediately after testing with the SSG. During the LWD test, the first three drops of the falling weight were applied as seating loads, followed by seven additional drops. The average LWD modulus was calculated based on the deflec- tions from the last three drops. Figure 3.28 shows the average modulus of the mixtures as measured with the LWD at 2 and 72 hours. (For clarity, error bars are not shown.) Overall, the LWD test was able to capture the effect of curing; however, three mixtures (Mixtures 7, 14, and 17) showed a lower modulus with respect to curing time. When individual specimens were examined, five had a lower modulus with respect to increased curing time with the LWD test. As with the SSG testing, results could be affected by the relatively small magnitude of the applied force and deformation levels in addition to testing variability (especially among replicates). Also, the zone of influence of the device is expected to be greater than the thickness of the test slabs. Table 3.17 shows the descriptive statistics of the LWD modulus with respect to curing time. From Table 3.17, the mean LWD stiffness increased with respect to curing time, as expected, showing the overall ability of the LWD test to capture the effect of curing. The spread of the LWD modulus was evaluated using the IQR. The IQR unexpectedly decreased slightly with respect to curing time. Table 3.18 shows the descriptive statistics of the LWD modulus with respect to recycling agent and presence of cement as an active filler. The mean LWD modulus increased when cement was included for both emulsified asphalt and foamed asphalt mixtures, as expected. Table 3.18 also shows that the IQR increased when cement was included for emulsi- fied asphalt mixtures but decreased when cement was included for foamed asphalt mixtures. This could suggest that the presence of cement as an active filler has a larger influence on the LWD modulus of mixtures containing emulsified asphalt. The within-specimen variability of the LWD modulus was also assessed via the COV. As shown in Figure 3.29, the COV for repetitive measurements was less than 10% except for a single data point. The average COV was 2.6% and 2.7% for the 2- and 72-hour testing, respectively. Replicate measure- ments at each curing time were possible for the SSG and LWD tests. The variability of the LWD modulus among mixture replicates is shown in Figure 3.30. Missing data points in the figure indicate no replicate specimen for a given mixture. As with the variability observed for the SSG stiffness, the COV among the mixtures for the LWD modulus was less than 30% except for Mixture 13 at 72 hours. The high variability Source DF f-Value p-Value Slab density 1 0.1 0.752 Recycling process 1 0.15 0.703 Recycling agent type 1 14.79 0.000 Curing time 1 14.05 0.000 Recycling agent content 6 3.22 0.005 Notes: DF = degrees of freedom; bolding indicates that the p-value shows the source to be significant. Table 3.16. Results of ANCOVA for SSG stiffness. 0 10 20 30 40 50 0 2 4 6 8 10 12 14 16 18 20 LW D M od ul us , k si Mixture ID 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.28. Modulus of mixtures as measured by LWD.

52 Curing Time Mean, ksi Minimum, ksi Quartile 1, ksi Quartile 3, ksi Maximum, ksi Range, ksi Interquartile Range, ksi 2 hours 18.8 4.6 11.8 23.4 31.0 26.4 11.6 72 hours 26.5 10.0 21.8 34.2 36.1 26.1 12.4 Table 3.17. Descriptive statistics of LWD modulus by curing time. Material Combination Mean, ksi Minimum, ksi Quartile 1, ksi Quartile 3, ksi Maximum, ksi Range, ksi Interquartile Range, ksi Emulsion, cement 25.6 4.6 9.3 35.8 36.1 31.5 26.5 Emulsion, no cement 21.3 9.7 13.6 25.6 33.1 23.5 12.0 Foam, cement 25.9 20.2 22.3 29.2 34.6 14.4 6.95 Foam, no cement 18.8 7.4 15.3 22.4 25.2 17.8 7.08 Table 3.18. Descriptive statistics of LWD modulus by recycling agent and active filler type. 0 5 10 15 20 25 0 4 8 12 16 20 24 28 32 C oe ff ic ie nt o f V ar ia tio n, % Specimens 2-hour 72-hour Figure 3.29. Within-specimen LWD variability in terms of a coefficient of variation. 0 10 20 30 40 50 0 2 4 6 8 10 12 14 16 18 20 C oe ff ic ie nt o f V ar ia tio n, % Mixture ID 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.30. LWD variability among mixture replicates in terms of a coefficient of variation.

53 with Mixture 13 was also observed for the SSG stiffness. The average COV was 14.9% and 15.2% for the 2- and the 72-hour tests, respectively. The curing ratio is shown in Figure 3.31. The curing ratio ranged from 0.9 to 7.8, with an average curing ratio of 1.8. Table 3.19 presents the ANCOVA results at a confidence level of 95%. The ANCOVA was performed to investigate statistically the effect of the recycled mixture parameters considered in this study on LWD modulus. Table 3.19 shows that the LWD modulus was significantly varied (the p-value was less than 0.05) as a function of curing time and recycling agent content with different recycling agent types, the same factors identified in the analysis of the SSG stiffness. The process type was not identified as a significant factor in the analysis of LWD modulus (the same observation as in SSG stiffness analysis). Unlike in the analysis of the SSG test, the density was identified as a significant factor for the mixtures considered in this study. The cement rate could not be esti- mated and was removed from the analysis by the statistical software (Minitab) for the same reasons cited in the SSG discussion. 3.4.3.3 Comparison Between SSG and LWD Tests The parameters used to assess the SSG and LWD test results were compiled and are shown in Table 3.20. This was done to rate the test methods to discern their ability to be used to make time-critical decisions regarding opening to traffic and surfacing of recycled materials. Using the column labeled “Desired Trend” as a guide, the range/observation for either the SSG or LWD was highlighted depending on which device better demonstrated the desired trend. As seen in Table 3.20, the LWD test generally identified the desired trend better than the SSG test based on the parameters and range of the mixtures used in this study. 3.4.3.4 Additional LWD Tests As discussed previously, additional sets of slabs were pre- pared for other tests, and the LWD testing was repeated on some of these additional slab sets at 1, 3, 6, and 24 hours of curing. The LWD test was performed immediately prior to the shear and raveling tests. Table 3.21 shows the details of the combined testing matrix from the long-pin shear test (LPST) and short-pin raveling test (SPRT) phases that resulted in a total of 18 mixtures for the additional LWD testing. Each mixture set included at least two replicates, and there were additional mixtures prepared at half the design binder con- tent of the original mixtures. This new set of LWD tests at curing times of 1, 3, 6, and 24 hours was analyzed, in a fashion similar to that used for the 2- and 72-hour LWD data, in order to investigate whether the LWD was able to provide consistent results (repeatability) and to capture the changes in material characteristics in a shorter curing duration. Figure 3.32 presents the LWD modulus at different curing times and for all material combinations considered. Fig- ure 3.32 indicates that curing time had an impact on the LWD modulus measured at a relatively shorter curing duration and more frequent intervals. (As presented previously, this was the case for the measurements collected at 2 and 72 hours.) Overall observations from Figure 3.32 and Table 3.22 include that the mean LWD modulus increased with an increase 0 2 4 6 8 10 0 2 4 6 8 10 12 14 16 18 20 C ur in g R at io Mixture ID Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.31. Curing ratio of mixtures from LWD. Source DF f-Value p-Value Slab density 1 10.98 0.001 Recycling process 1 0.11 0.745 Recycling agent type 1 3.9 0.080 Curing time 1 58.04 0.000 Recycling agent content 6 0.001 Notes: DF = degrees of freedom; bold/highlight = the p-value shows the source to be significant. 4.97 Table 3.19. Results of ANCOVA for LWD modulus.

54 Parameter Range/Observation Desired TrendSSG LWD Variability (within specimen) at 2 hours, COV 4% 2.6% Lower Variability (within specimen) at 72 hours, COV 6% 2.7% Lower Variability (among specimen replicates) at 2 hours, COV 14.3% 14.9% Lower Variability (among specimen replicates) at 72 hours, COV 18.2% 15.2% Lower Stiffness range at 2 hours 25.9 ksi 26.4 ksi Higher Stiffness range at 72 hours 21.2 ksi 26.1 ksi Higher Interquartile range at 2 hours 8.3 ksi 11.6 ksi Higher Interquartile range at 72 hours 7.4 ksi 12.4 ksi Higher Number of mixtures that lost stiffness over time 3 3 Lower Number of replicate specimens that lost stiffness over time 10 5 Lower Range of curing ratio (average) 0.69 to 5.36 (average = 1.44) 0.9 to 7.8 (average = 1.8) Higher Number of mixtures with statistically significant difference between the 2-hour vs. 72-hour tests 1 3 Higher Captured the effect of density Generally Generally Always Captured the effect of active filler presence Generally Generally Always Note: Highlights in columns denote which device better demonstrated the desired trend. Table 3.20. Comparison of SSG and LWD tests. Mix ID Stabilizing/Recycling Agent Active Filler Process State Agent Content, % Active Filler Content, % Actual Density, pcf No. of Replicates 1 Emulsified asphalt Cement CCPR IN 2.5 1.0 119.1 3 rep full0 rep half 2 VA 2.5 1.0 127.6 2 reps full0 rep half 3 FDR TX 0.5 1.1 131.5 2 rep full1 rep half 4 CA 2.5 1.0 127.8 3 reps full1 rep half 5 No cement CCPR NY 3.0 0.0 122.0 3 rep full0 rep half 6 VA 2.5 0.0 127.6 5 reps full0 rep half 7 CIR ON 1.2 0.0 121.4 2 rep full1 rep half 8 FDR IN 2.5 0.0 119.1 3 rep full0 rep half 9 CA 2.5 0.0 127.8 4 reps full0 rep half 10 Foamed asphalt Cement CCPR VA 2.5 1.0 127.6 5 reps full1 rep half 11 CIR CA 2.0 1.0 117.4 3 rep full1 rep half 12 MA 2.5 1.0 121.0 3 rep full1 rep half 13 FDR TX 2.4 1.5 125.6 4 reps full1 rep half 14 CA 2.5 1.0 127.8 3 reps full1 rep half 15 No cement CCPR VA 2.5 0.0 127.6 2 reps full1 rep half 16 CIR MI 2.2 0.0 129.8 3 rep full0 rep half 17 WI 2.0 0.0 121.3 3 rep full0 rep half 18 FDR CA 2.5 0.0 127.8 2 reps full1 rep half Table 3.21. Test slab details for LWD stiffness testing at 1, 3, 6, and 24 hours.

55 in curing time when the LWD moduli of all mixtures were combined without consideration of the specific characteristics of the mixtures. Likewise, the spread of the LWD modulus (as quantified by range and IQR), in general, also increased with an increase in curing time. Figure 3.32 also reveals that the impact of curing time on the LWD modulus is more evident with the presence of active filler (cement). Table 3.23 reflects that the spread of the LWD modulus is wider for the mixtures with cement than for the mixtures without cement. It was also noted from the table that the mixtures with cement tended to have a higher LWD modulus, as expected. Figure 3.32 reveals that there was, generally, a reduction in the measured LWD modulus followed by a slight increase in modulus over the four curing periods for the mixtures without cement, an observation also noted with other tests considered in this study. For those mixtures incorporating cement, in general, a continued increase in the LWD modulus with respect to an increase in curing time was observed, as expected. Figure 3.33 presents the mixture-to-mixture variability of the LWD modulus measured at shorter curing intervals in terms of the COV. The COV for the LWD measurements, considering all curing times, was less than 30% except for a few observations (4 of 72). It is interesting that for the curing time considered in this part of the study, two of four obser- vations with more than 30% COV were for Mixture 13. With regard to the LWD data for 2- and 72-hour tests in addition to observations from other tests used in this study, Mixture 13 had a consistently higher variability compared to the other mixtures. Nevertheless, the overall average COV for the LWD measurements at shorter curing intervals was 15.9%, which was slightly higher than the average COV of 14.9% and 15.2% 0 10 20 30 40 50 0 2 4 6 8 10 12 14 16 18 20 LW D M od ul us , k si Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, Cement Emulsion, No Cement Foam, No Cement Foam, Cement Figure 3.32. Modulus of mixtures as measured by LWD at shorter time intervals. Curing Time Mean, ksi Minimum, ksi Quartile 1, ksi Quartile 3, ksi Maximum, ksi Range, ksi Interquartile Range, ksi 1 hour 17.6 6.1 15.0 22.0 24.7 18.5 7.1 3 hours 18.2 8.7 14.7 22.4 28.5 19.8 7.7 6 hours 18.6 8.7 15.6 21.3 28.3 19.5 5.6 24 hours 22.2 11.4 18.6 26.6 34.3 22.8 8 Table 3.22. Descriptive statistics of LWD modulus by curing time. Material Combination Mean, ksi Minimum, ksi Quartile 1, ksi Quartile 3, ksi Maximum, ksi Range, ksi Interquartile Range, ksi Emulsion, cement 19.6 6.1 13.7 23.2 32.4 26.3 9.5 Emulsion, no cement 19.3 8.7 14.5 23.6 28.5 19.7 9.0 Foam, cement 19.4 14.5 15.9 21.9 34.3 19.8 6.0 Foam, no cement 18.3 14.6 16.1 20.9 22.1 7.5 4.8 Table 3.23. Descriptive statistics of LWD modulus by recycling agent and active filler type.

56 observed for the 2- and 72-hour LWD tests, respectively. The average COV for repetitive measurements was less than 1.5%, a statistic resulting from testing of 18 mixtures with at least two replicates at four different curing times with each test having three replicate measurements. The average COV for repeat measurements was 2.6% and 2.7% for the 2- and 72-hour LWD tests, respectively. Figure 3.33 shows that the magnitude of the observed variability does not vary as a function of the process type and recycling agent type or the curing time, the same observation as with the 2- and 72-hour LWD tests. As part of the study, selected mixtures (a total of 10) were purposely prepared with one-half binder content to inves- tigate whether the LWD test could capture such a change in material composition. Figure 3.34 presents the test results for those mixtures at curing times of 1, 3, 6, and 24 hours. Only one replicate was prepared for the half-binder content mixtures. The trend-wise performance behaviors of the half-binder mixtures were like those of the mixtures with the full binder content. In general, the LWD modulus increased with an increase in curing time for the mixtures with cement. There was an immediate reduction in the LWD modulus of the mixtures without cement, followed by an increase in modulus with increasing cure time. Similarly, the mixtures with cement tended to attain a higher LWD modulus. In comparing the magnitude of the LWD modulus of the mix- tures at each curing time, it was observed that the mixtures prepared with full binder content had a higher LWD modulus (in 31 of 40 observations) than the mixtures prepared with one-half binder content. These lower magnitudes of the 0 10 20 30 40 50 60 0 2 4 6 8 10 12 14 16 18 20 C oe ff ic ie nt o f V ar ia tio n, % Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, Cement Emulsion, No Cement Foam, No Cement Foam, Cement Figure 3.33. Variability of LWD modulus at shorter time intervals. 0 10 20 30 40 50 60 0 2 4 6 8 10 12 14 16 18 20 LW D S tif fn es s, ks i Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.34. LWD modulus of mixtures prepared at a half binder content tested at shorter time intervals.

57 LWD modulus were from Mixture 3 at all curing times, Mixture 12 at two curing times, Mixture 13 at a single curing time, and Mixture 14 at two curing times. It is interesting to note that the common factor for the mixtures was that they were prepared with cement, and three of the four mixtures were FDR mixtures. The test results presented herein show that overall the LWD test is sensitive to a change in a recycling agent content rate. Table 3.24 shows the ANCOVA results at a confidence level of 95% for the mixtures tested with an LWD at shorter curing intervals. This was the same analysis performed for the 2- and 72-hour LWD tests and combined all the mix- tures together to investigate statistically the effect of the recycled mixture parameters considered in this study on LWD modulus. It is evident from the table that the LWD modulus varied significantly (the p-value was less than 0.05) as a func- tion of curing time, recycling agent type, and process type and density. The results presented herein for shorter curing times indi- cated consistency with the results presented for the 2- and 72-hour LWD tests. However, in this analysis, the process type was identified as a statistically significant factor, potentially because of the increased number of mixtures, replicates, and curing time categories used in the analysis, which resulted in more data points and hence improved the power of the statistical analysis. The cement rate could not be estimated and was removed from the analysis by the statistical software (Minitab), as was the case previously described. 3.4.4 Deformation Resistance Using the same test slabs that were used for stiffness test- ing, deformation resistance testing was conducted using the upper assembly of an MH for 16 different mixtures fabricated from 12 sources, as shown in Table 3.12. For most of the mix- tures, tests were conducted on either two or three replicate slabs. In a limited number of cases, a single slab was tested because of the amount of material available. The MH test- ing was performed at two different locations (two corners) per slab at 2 hours and at the other two corners at 72 hours after compaction, leading to two average readings per curing dura- tion per fabricated slab. For each test, the penetrated depth was measured every five blows over a 20-blow test sequence. Figures 3.35 through 3.38 show the average penetrated depth at five, 10, 15, and 20 blows, respectively, for each mixture at 2 and 72 hours of curing. The reported values are the average of the penetrated depth at two locations per slab. These figures show that the test captured the effect of curing in that the average penetrated depth was less at 72 hours than at 2 hours for nearly all mixtures at all recorded blow counts. In addition, the difference in penetrated depth with respect to curing time increased as the number of blows increased. As expected, the penetrated depth increased from five to 10 blows, from 10 to 15 blows, and from 15 to 20 blows for all evaluated mixtures. Source DF f-Value p-Value Slab density 1 54.46 0.000 Recycling process 1 25.21 0.000 Recycling agent type 1 1.96 0.163 Curing time 1 9 0.000 Recycling agent content 6 9.25 0.000 Note: DF = degrees of freedom; bold/highlight = p -value shows the source to be significant. Table 3.24. Results of ANCOVA for LWD modulus at shorter curing times. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 1 2 3 5-1 5-2 6 7 8 11 12 13 14 16 17-1 17-2 18 A ve ra ge P en et ra te d D ep th , m m Mixture ID 5 Blows, 2 hrs Curing 5 Blows, 72 hrs Curing Figure 3.35. Average penetrated depths at 5 MH blows after 2- and 72-hour curing.

58 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 1 2 3 5-1 5-2 6 7 8 11 12 13 14 16 17-1 17-2 18 A ve ra ge P en et ra te d D ep th , m m Mixture ID 10 Blows, 2 hrs Curing 10 Blows, 72 hrs Curing Figure 3.36. Average penetrated depths at 10 MH blows after 2- and 72-hour curing. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 1 2 3 5-1 5-2 6 7 8 11 12 13 14 16 17-1 17-2 18 A ve ra ge P en et ra te d D ep th , m m Mixture ID 15 Blows, 2 hrs Curing 15 Blows, 72 hrs Curing Figure 3.37. Average penetrated depths at 15 MH blows after 2- and 72-hour curing. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 1 2 3 5-1 5-2 6 7 8 11 12 13 14 16 17-1 17-2 18 A ve ra ge P en et ra te d D ep th , m m Mixture ID 20 Blows, 2 hrs Curing 20 Blows, 72 hrs Curing Figure 3.38. Average penetrated depths at 20 MH blows after 2- and 72-hour curing.

59 Table 3.25 shows the descriptive statistics of the pene- trated depth with respect to number of MH blows (i.e., five, 10, 15, and 20) and curing time (i.e., 2 and 72 hours). The mean penetrated depth increased with respect to the number of blows and decreased with respect to curing time, as expected. Similarly, the IQR increased with respect to the number of blows and decreased with respect to curing time. The variability of the MH test results at the recorded blow counts was evaluated in terms of a COV computed for those mixtures that had replicates (11 of 16 mixtures). Figures 3.39 and 3.40 show the penetrated depth COV (averaged across multiple slabs) from all evaluated mixtures at 2 and 72 hours after fabrication, respectively, at five, 10, 15, and 20 MH blows. The COV decreased with an increasing number of blows. The research team attributed this to a reduced influ- ence of surface texture as the penetrated depth increased. COV values were lower for nearly all mixtures at 20 blows when compared with a lesser number of blows at both 2 and 72 hours. The average penetrated depth COV at five, 10, 15, and 20 blows was less at 2 hours after fabrication than at 72 hours after fabrication. At 2 hours after fabrication, seven, eight, and seven mixtures had COVs of less than 30% at 10, 15, and 20 blows, respectively. The generated data were further analyzed to investigate the effect of multiple parameters on the measured MH penetrated depth. An ANCOVA at a confidence level of 95% was used to evaluate the significance of these parameters. Table 3.26 presents the outcomes of the ANCOVA. The agent rate factor was nested within the recycling agent factor. Moreover, the cement content factor was nested within the recycling agent factor. The analysis shows that all factors, except the process type, significantly affected the MH pene- trated depth (i.e., p-value < 0.05). Number of Blows Curing Time Penetrated Depth, mm Mean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range 5 2 hours 1.22 0.46 0.66 1.44 3.33 2.87 0.7872 hours 0.58 0.26 0.39 0.76 1.19 0.93 0.37 10 2 hours 1.90 0.76 1.13 2.21 5.64 4.88 1.0872 hours 1.00 0.44 0.67 1.47 1.67 1.23 0.79 15 2 hours 2.49 1.03 1.54 3.00 6.81 5.78 1.4672 hours 1.28 0.53 0.96 1.72 2.27 1.74 0.76 20 2 hours 3.06 1.57 1.98 3.92 7.89 6.32 1.9472 hours 1.51 0.66 1.03 1.92 2.67 2.01 0.89 Table 3.25. Descriptive statistics of MH testing by number of blows and curing time. 0 10 20 30 40 50 60 70 80 90 100 1 2 6 7 8 11 12 13 14 16 18 C oe ff ic ie nt o f V ar ia tio n, % Mixture ID 5 blows - 2 hrs Curing 10 blows - 2 hrs Curing 15 blows - 2 hrs Curing 20 blows - 2 hrs Curing Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.39. Penetrated depth variability after 2-hour curing duration in terms of coefficient of variation.

60 3.4.5 Penetration Resistance Using the same test slabs used for testing deformation resistance, penetration resistance testing was conducted using a DCP for the 16 different mixtures fabricated from 12 sources (as shown in Table 3.12). For most of the mixtures, tests were conducted on either two or three replicate slabs. In a limited number of cases, a single slab was tested because of the amount of material available. The DCP test was con- ducted in accordance with ASTM D6951 with penetration readings collected after each blow. Figure 3.41 presents the DCP penetration index (DPI) for all evaluated mixtures. The DPI was calculated by dividing the total penetrated depth by the number of blows. The penetration rate was consistent throughout the depth of each slab, indicating material uniformity in the vertical direction. The data provided in Figure 3.41 show that the DPI ranged between 3.2 and 9.0 mm/blow after 2 hours of curing and between 1.2 and 7.3 mm/blow after 72 hours of curing. Table 3.27 shows the descriptive statistics of the DCP testing with respect to curing time. The mean DPI decreased with respect to curing time, as expected, showing the ability of the DCP test to capture the effect of curing. The spread of the DPI was evaluated using the IQR. Table 3.27 shows that the IQR decreased slightly with respect to curing time. Table 3.28 shows the descriptive statistics of the DPI testing with respect to recycling agent and presence of cement as an active filler. The mean DPI was decreased with respect to the presence of cement for both emulsified and foamed asphalt mixtures. The mean DPI was equal for emulsified and foamed asphalt mixtures containing cement but slightly lower for mixtures using foamed asphalt where no cement was present. The specimen-to-specimen variability was evaluated for the 11 mixtures having two or more replicates using the COV, as shown in Figure 3.42. The DPI at 2 hours showed a COV that ranged between 0.7% and 48.6%. Only three mixtures (Mixtures 1, 13, and 16) had a COV value greater than 10%. The DPI at 72 hours showed a COV that ranged between 0.9% and 23.6%. Only four mixtures (Mixtures 7, 13, 16, and 18) showed a COV value greater than 10%. Inter- estingly, Mixtures 13 and 16 showed COV values of greater than 10% after both curing times (i.e., 2 and 72 hours). It is unclear why the variability was much higher for these mixtures. The COV at 72 hours was less than the COV at 2 hours for about half the mixtures. The curing ratio, defined as the ratio of the DPI at 2 hours divided by the DPI at 72 hours, was used to evaluate the discrimination potential of DCP testing between curing times. Figure 3.43 shows the curing ratio of all evaluated mixtures. The computed ratio ranged between 1.24 and 5.13. No mix- tures exhibited a ratio lower than 1.0. An average ratio of 2.07 was calculated for all evaluated mixtures. 0 10 20 30 40 50 60 70 80 90 100 1 2 6 7 8 11 12 13 14 16 18 C oe ff ic ie nt o f V ar ia tio n, % Mixture ID 5 blows - 72 hrs Curing 10 blows - 72 hrs Curing 15 blows - 72 hrs Curing 20 blows - 72 hrs Curing Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.40. Penetrated depth variability after 72-hour curing duration in terms of coefficient of variation. Parameter DF f-Value p-Value Slab density 1 69.58 0.000 Recycling process 1 1.06 0.305 Recycling agent type 1 14.47 0.000 Curing time 1 129.20 0.000 Recycling agent content 6 15.84 0.000 Note: DF = degrees of freedom; bold/highlight = the p-value shows the source to be significant. Table 3.26. Marshall hammer testing: results of ANCOVA for MH penetration depth.

61 0 2 4 6 8 10 12 1 2 3 4 5-1 5-2 6 7 8 9 10 11 12 13 14 15 16 17-1 17-2 18 D PI , m m /b lo w 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Mixture ID Figure 3.41. DPI for all evaluated mixtures. 0 10 20 30 40 50 1 2 3 4 5-1 5-2 6 7 8 9 10 11 12 13 14 15 16 17-1 17-2 18 C oe ff ic ie nt o f V ar ia tio n, % Mixture ID 2-hour 72-hour Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.42. Coefficient of variation for DPI for mixtures with replicates. Curing Time DPI, mm/blowMean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range 2 hours 4.9 2.1 3.9 5.4 7.9 5.8 1.5 72 hours 2.7 1.0 1.9 3.4 4.8 3.8 1.4 Table 3.27. Descriptive statistics of DPI with respect to curing time. Material Combination DPI, mm/blow Mean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range Emulsion, cement 3.1 1.7 1.9 4.6 4.8 3.1 2.7 Emulsion, no cement 4.9 2.5 3.7 7.0 7.9 5.4 3.3 Foam, cement 3.1 1.0 2.0 4.2 5.6 4.6 2.2 Foam, no cement 4.3 2.7 3.5 5.1 6.0 3.3 1.6 Table 3.28. Descriptive statistics for DPI with respect to recycling agent and active filler type.

62 The 11 produced mixtures with two or more slab replicates were further evaluated to investigate the effect of multiple parameters on the DPI. These parameters included recycling agent content, recycling process, recycling agent type, curing time, and density. An ANCOVA at a confidence level of 95% was used to evaluate the significance of these parameters. The cement content factor was nested under the recycling agent factor. Table 3.29 presents the outcomes of the ANCOVA for DPI. The p-values in Table 3.29 show that the DPI was sensi- tive to all evaluated factors (p-value < 0.05) except the process type and density. A second round of DCP testing was conducted for selected mixtures using half the emulsified asphalt or foamed asphalt content to determine the influence of reducing the stabilizing/ recycling agent on the DPI. Although not shown here, DCP testing was conducted on Mixtures 3, 7, 11, 12, and 13 at 1 hour and 24 hours of curing at both the full and half binder contents. The results of this testing showed that the DPI was not sensitive to the reduction in binder content, suggesting that DCP testing may not be sensitive to parameters that could indicate a higher potential for raveling. 3.4.6 Shear Resistance An assessment of the shear resistance of the mixtures was made using a fixture developed in this study called a long-pin shear fixture. The LPST included an assessment of the number of blows required to drive the shear fixture into the test slab and then the maximum torque value reached prior to the material being sheared. A total of 18 mixtures were evaluated, as shown in Table 3.30. The collected data included the number of blows required to drive the shear fixture into the test slab and the maximum torque value obtained at 1, 3, 6, and 24 hours after compaction of the test slabs. Figures 3.44 and 3.45 show the average number of blows required to drive the shear fixture into the test slab and the average torque value at different curing times, respectively. Certain mixture numbers containing no data indicate those mixtures that were not fabricated for testing during this part of the study. Mixture numbers with missing data points indicate that the test was not performed at the designated curing time (e.g., the number of blow counts for Mixtures 4 and 9) or the test could not be performed because the mixture’s torque value reached the upper limit of the torque wrench (e.g., Mixtures 3 and 13 at 24 hours curing). It is evident from Figures 3.44 and 3.45 that the LPST measurements were affected by curing and the presence of cement. A curious trend was also observed for certain mix- tures in that the number of blows or torque value was seen to decrease from the 1- to the 3-hour test time and then increase at the 6- and 24-hour test times. The reasons for this are unclear, although this same trend was observed for certain mixtures during the field testing. 0 1 2 3 4 5 6 1 2 3 4 5-1 5-2 6 7 8 9 10 11 12 13 14 15 16 17-1 17-2 18 Pe ne tr at io n R at io Mixture ID Emulsion, Cement Emulsion, No Cement Foam, Cement Foam, No Cement Figure 3.43. Penetration ratio of all evaluated mixtures from DCP testing. Parameter DF f-Value p-Value Slab density 1 2.49 0.122 Recycling process 1 1.66 0.205 Recycling agent type 1 9.10 0.004 Curing time 1 58.38 0.000 Recycling agent content 4 4.33 0.005 Note: DF = degrees of freedom; bold/highlight = p -value shows the source to be significant. Table 3.29. Results of ANCOVA for DPI.

63 0 20 40 60 80 100 0 2 4 6 8 10 12 14 16 18 N um be r of B lo w s Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No CementFoam, Cement Emulsion, Cement Figure 3.44. Number of blows to drive shear fixture into laboratory produced slabs. 0 50 100 150 200 250 300 0 2 4 6 8 10 12 14 16 18 To rq ue , f t-l bs Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, Cement Emulsion, No Cement Foam, No Cement Foam, Cement Figure 3.45. Torque values for field long-pin shear test. Mix ID Stabilizing/Recycling Agent Active Filler Process State Agent Content, % Active Filler Content, % Actual Density, pcf No. of Replicates 1 Emulsified asphalt Cement CCPR IN 2.5 1.0 119.1 22 VA 2.5 1.0 127.6 0 3 FDR TX 4.5 1.1 131.5 24 CA 2.5 1.0 127.8 1 5 No cement CCPR NY 3.0 0.0 122.0 26 VA 2.5 0.0 127.6 3 7 CIR ON 1.2 0.0 121.4 2 8 FDR IN 2.5 0.0 119.1 29 CA 2.5 0.0 127.8 2 10 Foamed asphalt Cement CCPR VA 2.5 1.0 127.6 3 11 CIR CA 2.0 1.0 117.4 212 MA 2.5 1.0 121.0 2 13 FDR TX 2.4 1.5 125.6 214 CA 2.5 1.0 127.8 3 15 No cement CCPR VA 2.5 0.0 127.6 0 16 CIR MI 2.2 0.0 129.8 217 WI 2.0 0.0 121.3 2 18 FDR CA 2.5 0.0 127.8 0 Table 3.30. Test slab details for long-pin shear test at 1, 3, 6, and 24 hours.

64 The descriptive statistics of the LPST testing are shown in Tables 3.31 through 3.34. Tables 3.31 and 3.32 show the descriptive statistics of the number of blows by curing time and recycling agent type, respectively. Tables 3.33 and 3.34 show the descriptive statistics of the torque value by curing time and recycling agent type, respectively. Table 3.31 shows that the mean number of blows and the IQR increased with respect to curing time. Table 3.32 shows that the mean number of blows was greater when cement as an active filler was present for mixtures having both emulsified and foamed asphalt. The IQR was less when cement was included as an active filler for mixtures using emulsified asphalt but greater for mixtures using foamed asphalt. Table 3.33 shows that the mean torque value increased with respect to curing time, as expected. As noted previously, for some mixtures, the torque value at 24 hours could not be recorded since the value exceeded the maximum capacity of the handheld torque wrench. The IQR decreased from 1 to 3 hours but then increased from 3 to 24 hours. Table 3.34 shows that the torque values increased for mixtures using both emulsified and foamed asphalt with cement. The IQR was similar for emulsified mixtures with and without cement but decreased for foamed asphalt mixtures without cement. The variability of the LPST was assessed in terms of the COV calculated from testing replicate specimens. Figure 3.46 shows the COV for the number of blows, and Figure 3.47 shows the COV for the measured torque values. The COV for the number of blows was generally less than 20% (46 of 54 conditions, considering all curing times), with an average COV of 9.5%. Similarly, the measured torque value COV was generally less than 20% (47 of 49 conditions), with an average COV of 9%. Although some mixtures had a COV of greater than 20%, the data suggested that the variability was not par- ticularly affected by the process type, recycling agent type, or curing time. In general, those mixtures that had lower or higher COVs at the 1-hour test tended to have a relatively similar variability across all curing times. The generated data were also analyzed statistically to investigate the effect of the recycled mixture parameters considered in this study on the measured torque value and Curing Time Mean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range 1 hour 25.2 11 19.5 31.8 40 29 12.3 3 hours 29.5 13 21.3 35.6 50 37 14.5 6 hours 32.2 17 22.5 38.5 55 38 16.0 24 hours 41.5 19 26.0 52.5 80 61 26.5 Table 3.31. Descriptive statistics of number of blows by curing time. Material Combination Mean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range Emulsion, cement 40.1 19 33.3 44.0 72 53 10.8 Emulsion, no cement 25.0 11 18.0 31.5 40 29 13.5 Foam, cement 36.3 16 24.5 48.8 80 64 24.3 Foam, no cement 25.9 19 22.3 30.3 37 18 8.0 Table 3.32. Descriptive statistics of number of blows by recycling agent and active filler type. Curing Time Mean, ft-lbs Minimum, ft-lbs Quartile 1, ft-lbs Quartile 3, ft-lbs Maximum, ft-lbs Range, ft-lbs Interquartile Range, ft-lbs 1 hour 127.5 78.0 105.7 150.1 224.3 146.3 44.4 3 hours 128.6 68.7 110.5 149.8 212.7 144.0 39.3 6 hours 138.7 66.8 113.1 166.4 222.2 155.3 53.3 24 hours 147.5 76.8 114.4 179.3 217.3 140.6 64.8 Table 3.33. Descriptive statistics of torque values by curing time. Material Combination Mean, ft-lbs Minimum, ft-lbs Quartile 1, ft-lbs Quartile 3, ft-lbs Maximum, ft-lbs Range, ft-lbs Interquartile Range, ft-lbs Emulsion, cement 164.3 110.8 137.8 191.9 222.2 111.3 54.1 Emulsion, no cement 111.6 66.8 79.7 133.4 168.0 101.2 53.7 Foam, cement 151.7 94.7 116.1 182.6 224.3 129.7 66.5 Foam, no cement 131.2 99.8 115.4 146.2 179.3 30.879.5 Table 3.34. Descriptive statistics of torque values by recycling agent and active filler type.

65 the number of blows. An ANCOVA at a confidence level of 95% was used to test for significant factors in the number of blows and the torque among various mixture parameters. The factors used included process type, recycling agent type and rate, active filler (cement) rate, density, and curing time. The experiment was a nested design as it was not intended to have a factorial design for the levels of the various factors. In other words, although the recycling agent type (emul- sion or foam) was not nested as a factor in the process type (CIR or FDR), the recycling agent rate and cement rate were nested as a factor in the recycling agent type. Density was used as a covariate factor in the analysis. Table 3.35 shows the ANCOVA statistics for the number of blows. The number of blows was significantly varied (the p-value was less than 0.05) as a function of curing time, recycling agent rate with different recycling agent types, process, and density. Table 3.36 presents the ANCOVA sta- tistics for the torque value and shows that the torque values were significantly varied (the p-value was less than 0.05) as a function of curing time, recycling agent rate with different recycling agent types, process, and density. 3.4.7 Raveling Resistance Like the LPST, an assessment of the raveling resistance of the mixtures was made using a fixture developed in this study called a short-pin raveling fixture. The SPRT measured the same penetration and torque parameters as the LPST discussed in the previous section. A total of 18 mixtures were evaluated; mixture details are provided in Table 3.37. These mixtures were manufactured using 12 sources of recycled materials. For some mixtures, two slab replicates 0 10 20 30 40 0 2 4 6 8 10 12 14 16 18 C O V fo r N um be r of B lo w s, % Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.46. Variability of number of blows in terms of coefficient of variation. 0 10 20 30 40 0 2 4 6 8 10 12 14 16 18 C O V fo r To rq ue , % Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.47. Variability of torque value in terms of coefficient of variation.

66 Source DF f-Value p-Value Slab density 1 32.89 0.000 Recycling process 1 22.17 0.000 Recycling agent type 1 2.3 0.132 Curing time 3 20.38 0.000 Recycling agent content 6 0.000 Note: DF = degrees of freedom; bold/highlight = p -value shows the source to be significant. 10.38 Table 3.35. Results of ANCOVA for number of blows. Source DF f-Value p-Value Slab density 1 36.75 0.000 Recycling process 1 10.56 0.002 Recycling agent type 1 17.64 0.000 Curing time 3 6.44 0.001 Recycling agent content 6 0.000 Note: DF = degrees of freedom; bold/highlight = p -value shows the source to be significant. 5.57 Table 3.36. Results of ANCOVA for torque value. Mix ID Agent Filler Process State Agent Content, % Filler Content, % Actual Density, pcf No. of Replicates 1 Emulsified asphalt Cement CCPR IN 2.5 1.0 119.1 1 rep full0 rep half 2 VA 2.5 1.0 127.6 2 reps full0 rep half 3 FDR TX 4.5 1.1 131.5 1 rep full1 rep half 4 CA 2.5 1.0 127.8 2 reps full1 rep half 5 No cement CCPR NY 3.0 0.0 122.0 1 rep full0 rep half 6 VA 2.5 0.0 127.6 2 reps full0 rep half 7 CIR ON 1.2 0.0 121.4 1 rep full1 rep half 8 FDR IN 2.5 0.0 119.1 1 rep full0 rep half 9 CA 2.5 0.0 127.8 2 reps full0 rep half 10 Foamed asphalt Cement CCPR VA 2.5 1.0 127.6 2 reps full1 rep half 11 CIR CA 2.0 1.0 117.4 1 rep full1 rep half 12 MA 2.5 1.0 121.0 1 rep full1 rep half 13 FDR TX 2.4 1.5 125.6 2 reps full1 rep half 14 CA 2.5 1.0 127.8 2 reps full1 rep half 15 No cement CCPR VA 2.5 0.0 127.6 2 reps full1 rep half 16 CIR MI 2.2 0.0 129.8 1 rep full0 rep half 17 WI 2.0 0.0 121.3 1 rep full0 rep half 18 FDR CA 2.5 0.0 127.8 2 reps full1 rep half Table 3.37. Specimen and mixture details for short-pin raveling test. were produced, whereas one replicate was produced for other mixtures. Table 3.37 also shows that some replicates were produced using the full design stabilizing/recycling agent content, and some were produced at half the design content in an effort to force a more severe raveling situation. The raveling data are presented with respect to two mea- surements of the number of blows (N1 and N2) required to drive the raveling fixture into the test slab and the measured torque. The value N1 describes the number of blows required to drive the raveling fixture from the tip of the longer center pin to the tips of the shorter outer pins. The value N2 describes the number of blows required to drive the raveling fixture from the tip of the long center pin until the base plate was seated against the surface of the test slab. These two separate values were recorded in case they proved useful in the analysis. For each of these measurements, the data are presented at the full design binder content and for certain mixtures at half of the design binder content. Figures 3.48 and 3.49 show the number of blows (N1) at the full and half design binder contents, respectively. The

67 figures show that the mixtures with cement tended to have a greater magnitude and spread in the number of blows with respect to the four curing times than those mixtures with no cement. Table 3.38 shows the descriptive statistics for SPRT parameters using all collected data irrespective of curing time, recycling agent type, and cement content. Table 3.39 shows the descriptive statistics for the number of blows (N1) data with respect to curing time. The mean number of blows increased with respect to curing time. The IQR was similar for all four curing times. Table 3.40 shows the descriptive statistics for the number of blows (N1) with respect to recy- cling agent and cement content. The mean number of blows increased when cement was present in both emulsified and foamed asphalt mixtures. The IQR was similar for all material combinations. Figure 3.50 shows the variability of the number of blows (N1) in terms of the COV for those mixtures that had repli- cates. The COV values were all less than about 25% except for Mixture 13. There did not appear to be a clear trend with respect to material combination, in part because of the low number of mixtures that had replicates. Figures 3.51 and 3.52 show the number of blows (N2) required to drive the raveling fixture from the tip of the longer center pin until the base plate was seated against the surface of the test slab at the full and half binder contents, respectively. The mixtures with cement tended to have a greater magnitude and spread in the number of blows with respect to the four curing times than those mixtures that did not include cement. As with N1, N2 responded as expected to the presence of cement with respect to curing time. As 0 5 10 15 20 25 30 35 0 2 4 6 8 10 12 14 16 18 N 1, N um be r of B lo w s Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.48. Number of blows (N1) required to drive raveling fixture (full binder content). 0 5 10 15 20 25 30 35 0 2 4 6 8 10 12 14 16 18 N 1, n um be r of b lo w s Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.49. Number of blows (N1) required to drive raveling fixture (half binder content).

68 Parameter Mean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range Number of blows, N1 6.8 3.0 5.0 8.0 21.0 18.0 3.0 Number of blows, N2 12.1 6.0 9.0 14.0 34.0 28.0 5.0 Torque value 425.7 139.0 281.3 530.1 1,289 1,150 248.9 Table 3.38. Descriptive statistics of SPRT parameters for full and half binder specimens irrespective of curing time, recycling agent type, and cement content. Curing Time N1, BlowsMean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range 1 hour 5.3 3.0 4.0 7.0 8.0 5.0 3.0 3 hours 6.1 3.5 5.0 7.0 9.0 5.5 2.0 6 hours 7.1 3.5 5.0 8.0 13.5 10.0 3.0 24 hours 8.9 4.0 6.0 10.0 21.0 17.0 4.0 Table 3.39. Descriptive statistics of number of blows (N1) for full and half binder specimens with respect to curing time. Material Combination N1, Blows Mean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range Emulsion, cement 8.6 3.5 7.0 9.0 21.0 17.5 2.0 Emulsion, no cement 5.0 3.0 4.0 6.0 8.0 5.0 2.0 Foam, cement 7.7 4.0 5.0 9.0 16.0 12.0 4.0 Foam, no cement 5.4 3.5 5.0 6.0 9.0 5.5 1.0 Table 3.40. Descriptive statistics of number of blows (N1) for full and half binder specimens with respect to recycling agent type and cement content. 0% 20% 40% 60% 80% 100% 120% 0 2 4 6 8 10 12 14 16 18 C O V fo r N um be r of B lo w s ( N 1), % Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.50. Coefficient of variation for number of blows, N1 (full binder specimens only).

69 with N1, the range and magnitude of N2 were similar with respect to the use of emulsified asphalt or foamed asphalt. Comparing Figures 3.51 and 3.52, there did not appear to be a large difference between the number of blows (N2) between full and half binder contents. The descriptive statistics for N2 using all collected data irrespective of curing time, recycling agent type, and cement content are summarized in Table 3.41. The mean number of blows (N2) increased with respect to curing time, as expected. The IQR was similar for the first three curing times but generally increased with respect to curing time. Table 3.42 shows the descriptive statistics for N2 with respect to recycling agent and presence of cement. Table 3.42 shows that N2 increased when cement was present for mixtures using both emulsified and foamed asphalt. A similar trend was found for the IQR. When the results of N1 and N2 were compared, N2 showed a greater range with respect to curing time and material combinations and was thus a better descriptor than N1. Figure 3.53 shows the variability of N2 in terms of the COV for those mixtures that had replicates, with all values less than about 40% except for Mixture 4. There did not appear to be a clear trend based on material type, in part because of the low number of mixtures that had replicates. Figures 3.54 and 3.55 show the measured torque value using the raveling fixture at the full and half design binder contents, respectively. As with the number of blows, the torque values showed a greater magnitude and spread for those mixtures with cement. In addition, the magnitude and range for mixtures containing emulsified asphalt versus foamed asphalt were similar. As observed for the number of 0 5 10 15 20 25 30 35 0 2 4 6 8 10 12 14 16 18 N 2, n um be r of b lo w s Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.51. Number of blows (N2) required to drive raveling fixture (full binder content). 0 5 10 15 20 25 30 35 0 2 4 6 8 10 12 14 16 18 N 2, n um be r of b lo w s Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, CementEmulsion, Cement Figure 3.52. Number of blows (N2) required to drive raveling fixture (half binder content).

70 Curing Time N2, BlowsMean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range 1 hour 9.8 6.0 8.0 12.0 15.5 9.5 4.0 3 hours 11.2 7.0 9.0 13.0 16.0 9.0 4.0 6 hours 12.1 6.0 9.0 14.3 22.0 16.0 5.3 24 hours 15.7 8.0 10.0 17.0 34.0 26.0 7.0 Table 3.41. Descriptive statistics of number of blows (N2) for full and half binder specimens with respect to curing time. N2, BlowsMaterial Combination Mean Minimum Quartile 1 Quartile 3 Maximum Age Interquartile Range Emulsion, cement 15.0 7.5 11.8 16.3 34.0 26.5 4.5 Emulsion, no cement 9.5 6.5 8.0 11.3 14.0 7.5 3.3 Foam, cement 13.4 7.0 10.0 16.0 29.0 22.0 6.0 Foam, no cement 9.7 6.0 9.0 10.9 15.0 9.0 1.9 Table 3.42. Descriptive statistics of number of blows (N2) for full and half binder specimens with respect to recycling agent type and cement content. 0% 20% 40% 60% 80% 100% 0 2 4 6 8 10 12 14 16 18 C O V fo r N um be r of B lo w s ( N 2) , % Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.53. Coefficient of variation for number of blows, N2 (full binder specimens only). 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 16 18 To rq ue , f t-l bs Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No CementFoam, Cement Emulsion, Cement Figure 3.54. Torque value using raveling fixture (full binder content).

71 blows, Figure 3.55 shows that the torque values were similar at the half binder and at the full binder content. The descriptive statistics for the measured torque value using all collected data with respect to curing time are shown in Table 3.43. The mean torque value increased with respect to curing time. The IQR also increased with curing time from 1 to 6 hours but decreased from 6 to 24 hours. Table 3.44 shows the descriptive statistics for the measured torque value using all collected data with respect to material combina- tions. The mean torque value increased with the presence of cement for mixtures using both emulsified and foamed asphalt. A similar increasing trend was observed for the IQR. Figure 3.56 shows the variability of the torque value in terms of the COV. The COV values were considered low and were all less than about 25%. There did not appear to be a clear trend based on material type, partly because of the low number of mixtures that had replicates. An ANCOVA at a confidence level of 95% was used to evaluate the significance of various mixture parameters. The agent rate and cement type factors were both nested within the recycling agent type factor. Tables 3.45 through 3.47 present the outcomes of the ANCOVA for N1, N2, and ravel- ing torque, respectively. The p-values in Table 3.45 show that N1 was sensitive only to process type factor. Table 3.46 shows that the number of blows (N2) was sensitive to all evaluated factors. Table 3.47 shows that the measured torque value was sensitive to all factors except the recycling process and the recycling agent type. 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 16 18 To rq ue , f t-l bs Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No CementFoam, Cement Emulsion, Cement Figure 3.55. Torque value using raveling fixture (half binder content). Curing Time Raveling Torque, ft-lbMean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range 1 hour 29.0 15.0 21.0 34.7 59.3 44.3 13.7 3 hours 31.5 11.6 21.2 41.8 48.9 37.3 20.6 6 hours 36.1 16.3 23.3 47.6 65.9 49.5 24.3 24 hours 46.0 23.4 31.1 51.8 107.4 84.0 20.7 Table 3.43. Descriptive statistics of torque value using raveling fixture for full and half binder specimens with respect to curing time. Material Combination Raveling Torque, ft-lb Mean Minimum Quartile 1 Quartile 3 Maximum Range Interquartile Range Emulsion, cement 46.6 21.1 34.7 50.1 107.4 86.3 15.4 Emulsion, no cement 29.0 17.8 21.8 33.9 51.8 34.0 12.2 Foam, cement 37.9 15.0 25.5 48.7 78.3 63.3 23.3 Foam, no cement 26.4 11.6 20.4 34.0 44.4 32.8 13.6 Table 3.44. Descriptive statistics of torque value using raveling fixture for full and half binder specimens with respect to recycling agent and active filler type.

72 0% 20% 40% 60% 80% 100% 0 2 4 6 8 10 12 14 16 18 C O V fo r T or qu e, % Mixture ID 1-hour 3-hour 6-hour 24-hour Emulsion, No Cement Foam, No Cement Foam, Cement Emulsion, Cement Figure 3.56. Coefficient of variation for torque value (full binder specimens only). Parameter DF f-Value p-Value Slab density 1 24.09 0.000 Recycling process 1 6.37 0.014 Recycling agent type 1 4.14 0.046 Curing time 3 9.61 0.000 Note: DF = degrees of freedom; bolding indicates that the p-value shows the source to be significant. Table 3.45. Short-pin raveling test: results of ANCOVA for number of blows, N1. Parameter DF f-Value p-Value Slab density 1 33.81 0.000 Recycling process 1 20.29 0.000 Recycling agent type 1 10.37 0.002 Curing time 3 12.68 0.000 Note: DF = degrees of freedom; bolding indicates that the p-value shows the source to be significant. Table 3.46. Short-pin raveling test: results of ANCOVA for number of blows, N2. Parameter DF f-Value p-Value Slab density 1 34.08 0.000 Recycling process 1 0.69 0.411 Recycling agent type 1 4.26 0.000 Curing time 3 0.0025.43 Note: DF = degrees of freedom bolding indicates that the p-value shows the source to be significant. Table 3.47. Short-pin raveling test: results of ANCOVA for torque. 3.4.8 Correlation Analysis A correlation analysis was performed to investigate the relationship between selected test measurement combina- tions using the data from the Phase II laboratory study. The analysis was performed by calculating the Pearson correlation coefficient (r) and the associated p-value. The Pearson cor- relation coefficient describes the linear relationship between two variables and has a range of −1 < r < +1, where values closer to −1 or +1 indicate a stronger correlation. A value of −1 or +1 indicates a negative or positive relationship, respec- tively. The p-value indicates the statistical significance of the relationship; a higher p-value suggests that the correlation may be due to random chance. Tables 3.48 and 3.49 show the Pearson correlation coef- ficient and p-value for comparisons of the test slab density, SSG stiffness, LWD modulus, and DPI values at curing periods of 2 and 72 hours, respectively. For those combinations that were shown to have a strong correlation (|r| > 0.7, based on categories by Evans [1996]), the p-value was determined to estimate the significance of the relationship (alpha = 0.05). Shaded cells indicate comparisons where both conditions were met. There was a strong correlation between the SSG and LWD at 2 hours of curing. For both comparisons, the p-value indicates that the correlation was statistically significant. Tables 3.50 through 3.53 show the results of the correlation analysis for those slabs tested at 1, 3, 6, and 24 hours of curing time, respectively. The tables show the Pearson cor- relation coefficient and p-value for comparisons of the test slab density, LWD modulus, LPST number of blows, LPST torque value, SPRT number of blows (N1 and N2), SPRT torque value, and DPI values. Results from MH testing were

73 not included because of the high test variability. For those combinations that were shown to have a strong correlation (|r| > 0.7), the p-value was determined to estimate the signifi- cance of the relationship (alpha = 0.05). Shaded cells indicate comparisons where both conditions were met. The number of blows from the LPST and SPRT have a strong correlation, and the relationship was statistically signi- ficant across all four curing times. The LPST torque value and the SPRT number of blows also had a strong correlation with the DPI, with relationships statistically significant at 1 and 24 hours of curing. The SPRT torque value had a strong correlation with the DPI but with the relationship statistically significant only at the 24-hour curing time. Slab density did not have a strong correlation with any of the performance tests. The LPST torque value did not have a strong correlation with the SPRT torque value. Figures 3.57 through 3.60 demonstrate the relationship between those tests shown in the correlation analysis to have the strongest correlation and a statistically significant relationship. The data are presented with respect to the curing time, and linear trendlines are shown for each. Another trendline type (e.g., polynomial, exponential) might show a higher coefficient of determination, but use of a linear trend is consistent with the relationship shown by the Pearson correlation coefficient. Figures 3.57 and 3.58 show the relationship between the LPST number of blows and the SPRT number of blows N1 and N2, respectively, for curing times of 1, 3, 6, and 24 hours. The trendline slopes were similar for all curing times, and the coefficient of determination generally increased with respect to curing time. Figure 3.59 shows the relationship between the SPRT number of blows N1 and N2 for all curing times. The trendline slope values were similar across all curing times, and the coefficient of determination increased with respect to curing time. Figure 3.60 shows the relation- ship between the LPST torque value and the SPRT number of blows (N2) for all curing times. The trendline slopes were similar for all curing times, and the coefficient of determina- tion generally increased with respect to curing time. Figure 3.61 shows the relationship between the LPST torque value and DPI for curing times of 1 and 24 hours. The coeffi- cient of determination increased with respect to curing time, and the slope of the trendline became more negative as curing time increased. Figure 3.62 shows the relationship between the SPRT torque value and DPI for both curing times. As with Figure 3.61, the coefficient of determination increased with respect to curing time, but the slope of the trendline became less negative as curing time increased. Figure 3.63 shows the relationship between the SPRT N2 and DPI for both curing times. The coefficient of determination increased slightly with respect to curing time, and the slope of the trendline became less negative as curing time increased. A nonlinear trendline would likely better describe the relationships shown in these three figures. SSG Stiffness, MN/m LWD Modulus, ksi DPI, mm/blow Slab density, lb/ft3 −0.1815 0.4664 0.0461 SSG stiffness, MN/m 0.7472 −0.6572 LWD modulus, ksi −0.5539 (b) SSG Stiffness, MN/m LWD Modulus, ksi DPI, mm/blow Slab density, lb/ft3 0.3370 0.0094 0.8089 SSG stiffness, MN/m 0.0000 0.0001 LWD modulus, ksi 0.0015 (a) Table 3.48. Correlation analysis at 2 Hours of curing, (a) Pearson correlation coefficient, (b) p-value. (a) SSG Stiffness, MN/m LWD Modulus, ksi DPI, mm/blow Slab density, lb/ft3 −0.0178 −0.1538 −0.1976 SSG stiffness, MN/m 0.2105 −0.2108 LWD modulus, ksi −0.5149 (b) SSG Stiffness, MN/m LWD Modulus, ksi DPI, mm/blow Slab density, lb/ft3 0.9258 0.4170 0.2952 SSG stiffness, MN/m 0.2641 0.2636 LWD modulus, ksi 0.0036 Table 3.49. Correlation analysis at 72 hours of curing, (a) Pearson correlation coefficient, (b) p-value.

74 (a) LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 −0.5056 −0.2964 0.1018 −0.0381 −0.0181 0.3018 −0.5755 LWD modulus, ksi 0.4620 0.0167 0.4550 0.5461 0.3165 * LPST number of blows 0.5296 0.7690 0.7484 0.1404 −0.6922 LPST torque value, ft-lb 0.7659 0.8595 0.5736 −0.7109 SPRT number of blows, N1 0.9475 0.3410 −0.9167 SPRT number of blows, N2 0.4750 −0.8611 SPRT torque value, ft-lb −0.6323 * = Combination assessed as part of 2 and 72 hours comparison. (b) LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 0.0032 0.0995 0.5792 0.8227 0.9154 0.0695 0.0197 LWD modulus, ksi 0.0078 0.9276 0.0047 0.0005 0.0563 * LPST number of blows 0.0136 0.0000 0.0001 0.5438 0.0183 LPST torque value, ft-lb 0.0014 0.0000 0.0066 0.0142 SPRT number of blows, N1 0.0000 0.1630 0.0000 SPRT number of blows, N2 0.0296 0.0000 SPRT torque value, ft-lb 0.0086 * = Combination assessed as part of 2 and 72 hours comparison. Table 3.50. Correlation analysis at 1-hour of curing, (a) Pearson correlation coefficient, (b) p-value.

75 (a) LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 −0.5269 0.0000 0.0152 0.2209 0.2335 0.4979 DNT LWD modulus, ksi 0.2978 0.2573 0.5365 0.6171 0.1631 * LPST number of blows 0.8734 0.8663 0.8315 −0.0319 DNT LPST torque value, ft-lb 0.9048 0.8644 −0.1108 SPRT number of blows, N1 0.9565 0.1154 SPRT number of blows, N2 0.0916 SPRT torque value, ft-lb DNT = did not test; * = combination assessed as part of 2 and 72 hours comparison. (b) LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 0.0019 0.0978 0.9340 0.2093 0.1838 0.0027 DNT LWD modulus, ksi 0.9998 0.0170 0.0011 0.0001 0.3566 * LPST number of blows 0.0000 0.0000 0.0000 0.8969 DNT LPST torque value, ft-lb 0.0001 0.0000 0.6515 SPRT number of blows, N1 0.0000 0.6381 SPRT number of blows, N2 0.7091 SPRT torque value, ft-lb DNT = did not test; * = combination assessed as part of 2 and 72 hours comparison. Table 3.51. Correlation analysis at 3 hours of curing, (a) Pearson correlation coefficient, (b) p-value.

76 (a) LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 −0.4124 0.1273 0.2375 0.1892 0.1677 0.4711 DNT LWD modulus, ksi 0.4459 0.3251 0.7499 0.7673 0.5465 * LPST number of blows 0.8050 0.7906 0.8526 0.5575 DNT LPST torque value, ft-lb 0.8217 0.8090 0.5099 SPRT number of blows, N1 0.9746 0.6258 SPRT number of blows, N2 0.6688 SPRT torque value, ft-lb DNT = did not test; * = combination assessed as part of 2 and 72 hours comparison. (b) LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 0.0190 0.4873 0.1905 0.2620 0.3212 0.0032 DNT LWD modulus, ksi 0.0105 0.0694 0.0000 0.0000 0.0005 * LPST number of blows 0.0000 0.0000 0.0000 0.0086 DNT LPST torque value, ft-lb 0.0003 0.0000 0.0182 SPRT number of blows, N1 0.0000 0.0024 SPRT number of blows, N2 0.0009 SPRT torque value, ft-lb DNT = did not test; * = combination assessed as part of 2 and 72 hours comparison. Table 3.52. Correlation analysis at 6 hours of curing, (a) Pearson correlation coefficient, (b) p-value.

77 (a) LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 −0.2052 0.2708 0.1729 0.2712 0.2887 0.4517 −0.5248 LWD modulus, ksi 0.6020 0.4114 0.8302 0.8780 0.6623 * LPST number of blows 0.8479 0.9164 0.9405 0.2967 −0.6440 LPST torque value, ft-lb 0.8555 0.9163 0.3874 −0.9292 SPRT number of blows, N1 0.9756 0.4126 −0.8860 SPRT number of blows, N2 0.4371 −0.8733 SPRT torque value, ft-lb −0.8755 * = Combination assessed as part of 2 and 72 hours comparison. LWD Modulus, ksi LPST Number of Blows LPST Torque Value, ft-lb SPRT Number of Blows, N1 SPRT Number of Blows, N2 SPRT Torque Value, ft-lb DPI, mm/blow Slab density, lb/ft3 0.2599 0.1338 0.3440 0.1268 0.1032 0.0083 0.0446 LWD modulus, ksi 0.0003 0.0296 0.0000 0.0000 0.0000 * LPST number of blows 0.0000 0.0000 0.0000 0.2475 0.1185 LPST torque value, ft-lb 0.0016 0.0000 0.1244 0.0025 SPRT number of blows, N1 0.0000 0.0998 0.0000 SPRT number of blows, N2 0.0794 0.0000 SPRT torque value, ft-lb 0.0000 * = Combination assessed as part of 2 and 72 hours comparison. (b) Table 3.53. Correlation analysis at 24 hours of curing, (a) Pearson correlation coefficient, (b) p-value.

78 y = 0.1647x + 1.4610 R² = 0.5914 y = 0.1495x + 2.0120 R² = 0.7505 y = 0.2064x + 0.7044 R² = 0.625 y = 0.1844x + 1.2269 R² = 0.8398 0 10 20 30 0 10 20 30 40 50 60 70 SP R T N um be r of B lo w s, N 1 LPST Number of Blows 1-hour 3-hour 6-hour 24-hour 1-hour Trendline 3-hour Trendline 6-hour Trendline 24-hour Trendline Figure 3.57. Relationship between LPST number of blows and SPRT N1. y = 0.2688x + 3.6838 R² = 0.5600 y = 0.2270x + 4.8617 R² = 0.6913 y = 0.3236x + 2.3154 R² = 0.727 y = 0.3271x + 2.4185 R² = 0.8846 0 10 20 30 40 0 10 20 30 40 50 60 70 SP R T N um be r of B lo w s, N 2 LPST Number of Blows 1-hour 3-hour 6-hour 24-hour 1-hour Trendline 3-hour Trendline 6-hour Trendline 24-hour Trendline Figure 3.58. Relationship between LPST number of blows and SPRT N2. y = 1.5894x + 1.5340 R² = 0.8978 y = 1.5135x + 1.8378 R² = 0.9149 y = 1.4167x + 2.3333 R² = 0.9498 y = 1.6859x + 0.9283 R² = 0.9519 0 10 20 30 40 0 5 10 15 20 SP R T N um be r of B lo w s, N 2 SPRT Number of Blows, N1 1-hour 3-hour 6-hour 24-hour 1-hour Trendline 3-hour Trendline 6-hour Trendline 24-hour Trendline Figure 3.59. Relationship between SPRT N1 and SPRT N2.

79 y = 0.0685x + 1.3869 R² = 0.7387 y = 0.0675x + 2.8250 R² = 0.7471 y = 0.0933x - 1.0626 R² = 0.6545 y = 0.0943x - 0.4478 R² = 0.8397 0 10 20 30 40 0 50 100 150 200 250 SP R T N um be r of B lo w s, N 2 LPST Torque, ft-lb 1-hour 3-hour 6-hour 24-hour 1-hour Trendline 3-hour Trendline 6-hour Trendline 24-hour Trendline Figure 3.60. Relationship between LPST torque value and SPRT N2. y = -0.0366x + 11.0864 R² = 0.5053 y = -0.0691x + 15.5393 R² = 0.8635 0 2 4 6 8 10 12 14 50 70 90 110 130 150 170 190 210 230 250 LPST Torque, ft-lb 1-hour 24-hour 1-hour Trendline 24-hour Trendline D PI , m m /b lo w Figure 3.61. Relationship between LPST torque value and DPI. y = -0.0910x + 9.1410 R² = 0.3998 y = -0.0709x + 8.2700 R² = 0.7665 0 2 4 6 8 10 12 14 0 20 40 60 80 100 120 D PI , m m /b lo w SPRT Torque, ft-lb 1-hour 24-hour 1-hour Trendline 24-hour Trendline Figure 3.62. Relationship between SPRT torque value and DPI.

80 Figures 3.64 and 3.65 show the relationship between the LWD modulus and the SPRT N1 and N2, respectively, at curing times of 1, 3, 6, and 24 hours. From both figures, the explanation of the relationship with a linear trend was generally poor at early curing times but improved (i.e., the coefficient of determination increased) as curing time pro- gressed. It is not clear if this is an indication of issues with conducting the test on the slab surface at early curing times or another phenomenon. The slope of the trendline increased with respect to curing time in both figures. 3.4.9 Proposed Tests for Field Testing Following the laboratory testing, selected tests were suggested for field testing based on the tests’ ability to provide responses with low variability and sensitivity with respect to curing time, recycling agent content, and the use of an active filler. Based on the laboratory experiment, these factors were qualitatively evaluated for each test and given a score of good, fair, or poor, as shown in Table 3.54. For variability, the COV was evaluated, and test results were assigned a rating of good, fair, or poor when the between-specimen COV was less than 20%, between 20% and 30%, or greater than 30%, respectively. For range, the ANCOVA results were reviewed, and test results were assigned a rating of good, fair, or poor if the p-value was less than 0.001, 0.001 to 0.05, or greater than 0.05, respectively. All tests except the MH test were recommended for field testing. The MH test was not recommended for three primary reasons. First, the test had a high variability. Second, y = -0.6987x + 14.0290 R² = 0.7415 y = -0.2370x + 8.9614 R² = 0.7627 0 2 4 6 8 10 12 0 5 10 15 20 25 30 35 40 D PI , m m /b lo w SPRT Number of Blows N2 1-hour 24-hour 1-hour Trendline 24-hour Trendline Figure 3.63. Relationship between SPRT N2 and DPI. y = 0.1253x + 2.9528 R² = 0.2071 y = 0.1863x + 2.6591 R² = 0.2878 y = 0.343x + 0.5406 R² = 0.5624 y = 0.3754x - 0.5557 R² = 0.6892 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 SP R T N um be r of B lo w s N 1 LWD Modulus, ksi 1-hour 3-hour 6-hour 24-hour 1-hour Trendline 3-hour Trendline 6-hour Trendline 24-hour Trendline Figure 3.64. Relationship between LWD modulus and SPRT N1.

81 the statistical analysis showed that the test was less sensitive to key material factors, including presence of active filler and recycling process. Third, and although not assessed directly, the depth of penetration into the recycled material was also small compared to the particle size, and so changes in surface texture may have had a greater effect on the results of the test. When the number of blows for the SPRT were counted, there was very little statistical difference between choosing the number of blows to drive the fixture from the tip of the center pin to the tip of the outer pins (N1) versus driving the fixture the full length of the center pin (N2). However, practically, it is easier to assess when the fixture base plate is flush with the surface of the recycled layer than to determine when the tips of the outer pins are touching. For field testing, counting the number of blows to drive the full length of the center pin (N2) is recommended. 3.4.10 Ruggedness Evaluation A ruggedness evaluation was completed to assess the impact of varying the operating conditions and equipment tolerances for both the LPST and SPRT (torque and number of blows for each). For both tests, factors (and levels within these factors) that were expected to influence the test results were varied, and the resulting test value was analyzed with respect to the variation. Tables 3.55 and 3.56 show results y = 0.2531x + 5.1593 R² = 0.2982 y = 0.3485x + 4.4760 R² = 0.3808 y = 0.5414x + 1.9538 R² = 0.5888 y = 0.6477x - 0.9824 R² = 0.7708 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 SP R T N um be r of B lo w s N 2 LWD Modulus, ksi 1-hour 3-hour 6-hour 24-hour 1-hour Trendline 3-hour Trendline 6-hour Trendline 24-hour Trendline Figure 3.65. Relationship between LWD modulus and SPRT N2. Property Test Criterion Between- Specimen Variability Range Other Test(s) Having Strong CorrelationCuring Time Recycling Agent Content Stiffness Soil stiffness gauge Fair Good Fair LWD, DPI Lightweight deflectometer Fair Good Fair SSG Deformation resistance Marshall hammer Poor Good Good Not assessed Penetration resistance Dynamic cone penetrometer Good Good Good LPST torque, SPRT number of blows and torque Shear resistance Long-pin shear test (blows/torque) Fair/good Good/fair Good/good LPST number of blows and torque, SPRT number of blows and torque, DPI Raveling resistance Short-pin raveling test (blows/torque) Good/good Good/fair Good/poor LPST number of blows and torque, SPRT number of blows, DPI Table 3.54. Assessment of tests for field testing recommendation.

82 of testing along with the factors and levels for the LPST and SPRT, respectively. The factors and values for each level were determined based on the results of concurrent laboratory testing, limited field testing, and the engineering judgment of the research team. In accordance with ASTM C1067, statistical parameters were calculated to identify which factors significantly influ- enced the test results. The factor effects were estimated by calculating the difference between average results at the high (+1) and low (−1) levels. The half-normal plots of the effects on number of blows and torque measurements for LPST and SPRT are shown in Figures 3.66 and 3.67, respec- tively. The data points that are farthest to the right of the reference line are potentially significant factors. Student’s t-tests at a 5% significance level were performed to the factors that significantly influenced the test results, and the results of the analysis are tabulated in Table 3.57 for both tests. The outer pin diameter was identified as a significant factor for the LPST number of blows, and the outer pin diameter and torque angular rate were identified as signifi- cant for the LPST torque value. The pin length was identified as a significant factor for the SPRT number of blows, and the tip dullness was identified as significant for the SPRT torque value. 3.5 Field Testing Based on the results of the laboratory testing, field testing was completed at nine different projects during the 2019 construction season. The field projects included CIR, CCPR, and FDR using either emulsified or foamed asphalt as the stabilizing/recycling agent with and without cement as an active filler. The projects were completed by multiple contractors, using different source materials, and were in different climatic regions. Table 3.58 shows the tests that were conducted to assess the desired properties. The following sections show the results of the data collec- tion at the nine field project sites. Given the objectives of the study, only those data collected during the first 3 hours of curing are shown, although, in some cases, data were collected for up to 24 hours of curing. For each figure, error bars show plus/minus one standard deviation calculated from replicate test blocks. The numerical value at the base of each column indicates the number of test blocks that were tested. Specimen No. Factor Result Pin Length, in. Tip Angle, ° Torque Angular Rate, °/sec Tip Dullness Outer Pin Diameter, in. Number of Blows Torque, ft-lb 1 3.1 85 90 Dull 13/32 13 46.4 2 2.9 85 90 Sharp 1/2 14 31.8 3 2.9 65 90 Sharp 13/32 14 47.7 4 3.1 65 60 Sharp 1/2 15 36.6 5 2.9 85 60 Dull 1/2 16 34.4 6 3.1 65 90 Dull 1/2 16 37.1 7 3.1 85 60 Sharp 13/32 12 30.0 8 2.9 65 60 Dull 13/32 12 38.9 Table 3.55. Long-pin shear test ruggedness factors and results. Specimen No. Factor Result Pin Length, in. Tip Angle, ° Torque Angular Rate, °/sec Tip Dullness Outer Pin Diameter, in. Number of Blows Torque, ft-lb 1 0.85 70 90 Dull 13/32 5 17.2 2 0.65 70 60 Sharp 1/2 4 12.3 3 0.65 50 90 Sharp 13/32 4 17.7 4 0.85 50 60 Sharp 1/2 5 13.8 5 0.65 70 60 Dull 1/2 5 17.2 6 0.85 50 90 Dull 1/2 5 17.9 7 0.85 70 60 Sharp 13/32 5 12.8 8 0.65 50 60 Dull 13/32 4 13.5 Table 3.56. Short-pin raveling test ruggedness factors and results.

83 (a) (b) Pin Length Torque Angular Rate Tip Angle Outer Pin Diameter Tip DullnessReference Line with Slope 1/Seffect 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 H al f N or m al Effect, Number of Blows Torque Angular Rate Tip Dullness Pin Length Outer Pin Diameter Tip AngleReference Line with Slope 1/Seffect 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.0 2.0 4.0 6.0 8.0 10.0 H al f N or m al Effect, Torque Figure 3.66. Half-normal plot for LPST, (a) number of blows, (b) torque value.

84 (a) (b) Outer Pin Diameter Torque Angular Rate Tip Angle Tip Dullness Length Reference Line with Slope 1/Seffect 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 H al f N or m al Effect, Number of Blows Outer Pin Diameter Pin Length Tip Angle Torque Angular Rate Tip Dullness Reference Line with Slope 1/Seffect 0 0.5 1 1.5 2 2.5 0.0 0.5 1.0 1.5 2.0 2.5 H al f N or m al Effect, Torque Figure 3.67. Half-normal plot for SPRT, (a) number of blows, (b) torque value. Test PropertyMeasure Factor Pin Length, in. Tip Angle, ° Torque Angular Rate, °/sec Tip Dullness Outer Pin Diameter, in. Long-pin shear test No. of blows NS NS NA NS S Torque, ft-lb NS NS S NS S Short-pin raveling test No. of blows S NS NA NS NS Torque, ft-lb NS NS NS S NS NS = not significant; S = significant; NA = not applicable. Table 3.57. Statistical significance of factors for LPST and SPRT. Property Test Density Nuclear gauge density Stiffness Soil stiffness gauge Lightweight deflectometer Penetration resistance Dynamic cone penetrometer Shear resistance Long-pin shear test (number of blows and torque value) Raveling resistance Short-pin raveling test (number of blows [N2] and torque value) Table 3.58. Properties assessed and tests conducted during field testing.

85 3.5.1 Stiffness 3.5.1.1a Soil Stiffness Gauge Figure 3.68 shows the results of field testing using the SSG for FDR and CCPR mixtures, and Figure 3.69 shows them for CIR mixtures. Figure 3.68 shows that the SSG stiffness of two of the three FDR mixtures and both CCPR mixtures was similar. The SSG stiffness of the FDR in Cell 1 from Minnesota was less than that of the other FDR and CCPR mixtures. Given that the SSG measurement zone extends beyond the depth of the recycled layer, this test result is likely influenced by the underlying support condition. When Figure 3.69 is compared with Figure 3.68, the SSG stiffness of the CIR mixtures is similar to that of the FDR and CCPR mixtures. 3.5.1.2 Lightweight Deflectometer Figure 3.70 shows the LWD modulus field testing results for FDR and CCPR mixtures, and Figure 3.71 shows them for CIR mixtures. Figure 3.70 shows that the FDR and CCPR mixtures had similar LWD modulus values except for the FDR sections from Minnesota. This same trend was observed for the SSG stiffness values. The LWD measurement zone is also known to extend beyond the depth of the recycled layer, and so the underlying foundation is likely influencing this test result. 2 2 3 5 3 0 5 10 15 20 25 30 35 40 1 hour 1 hour 3 hours 3 hours 1 hour 1 hour SC 123 NM US 491 MN Cell 1-HD MN Cell 1-LD MN Cell 7 MN Cell 5 FDR F-C FDR F-C FDR E-C FDR E-C CCPR F-C CCPR E-N SS G S tif fn es s, M N /m Curing Time; Project; Recycling Process, Agent, and Active Filler Figure 3.68. Soil stiffness gauge field testing results, FDR and CCPR mixtures. 1 1 3 1 2 1 3 2 0 5 10 15 20 25 30 35 40 1 hour 1.5 hours 1 hour 2 hours 1 hour 1 hour 1 hour 1 hour 1 hour 1 hour NY 23A NY 23A MN Cell 3 CA SR 22 NY 30 NY 28 MN Cell 4 IN SR 1- GS IN SR 1- PS CA SR 178 CIR F-N CIR F-N CIR F-C CIR F-C CIR E-N CIR E-N CIR E-N CIR E-N CIR E-N CIR E-C SS G S tif fn es s, M N /m Curing Time; Project; Recycling Process, Agent, and Active Filler Figure 3.69. Soil stiffness gauge field testing results, CIR mixtures.

86 Figure 3.71 shows that the LWD modulus of the CIR projects was like the FDR and CCPR modulus values. How- ever, two CIR projects showed much higher values, and one showed much lower values, than the other projects. Two of these relatively extreme values occurred from two test sections at the project from Indiana. This project was constructed on a section of roadway that had good foundation material (shown in Figure 3.71 as GS for good support) in the travel lanes but poor quality material (shown in Figure 3.71 as PS for poor support) in the shoulder areas. The research team intentionally tested in these two locations to give a wider range of material properties. Figure 3.71 shows that the support conditions likely influenced the test results. As can be observed in the results for the other tests, those tests that act only on the recycled layer do not show the same differ- ence in properties as identified by the LWD modulus for the GS and PS sections at the Indiana project. 3.5.2 Penetration Resistance Figure 3.72 shows the results of DCP field testing for FDR and CCPR mixtures, and Figure 3.73 shows them for CIR 2 2 3 1 5 3 0 5 10 15 20 25 30 35 40 45 1 hour 1 hour 3 hours 3 hours 1 hour 1 hour SC 123 NM US 491 MN Cell 1-HD MN Cell 1-LD MN Cell 7 MN Cell 5 FDR F-C FDR F-C FDR E-C FDR E-C CCPR F-C CCPR E-N LW D M od ul us , k si Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-C = foam plus cement; E-C = emulsion plus cement; E-N = emulsion, no cement. Figure 3.70. LWD modulus field testing results, FDR and CCPR mixtures. 1 1 3 1 2 1 3 1 2 2 0 5 10 15 20 25 30 35 40 45 1 hour 1.5 hours 1 hour 2 hours 1 hour 1 hour 1 hour 1 hour 1 hour 1 hour NY 23A NY 23A MN Cell 3 CA SR 22 NY 30 NY 28 MN Cell 4 IN SR 1- GS IN SR 1- PS CA SR 178 CIR F-N CIR F-N CIR F-C CIR F-C CIR E-N CIR E-N CIR E-N CIR E-N CIR E-N CIR E-C LW D M od ul us , k si Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-N = foam, no cement; F-C = foam plus cement; E-C = emulsion plus cement; E-N = emulsion, no cement. Figure 3.71. LWD modulus field testing results, CIR mixtures.

87 mixtures. The DCP penetration index values are similar for the FDR and CCPR projects except for the FDR section from New Mexico. Despite the low stiffness and modulus values indicated earlier, the Minnesota FDR sections do not show a significant decrease in the penetration index. The Minnesota FDR section with a lower density (noted as LD in Figure 3.72) had a higher penetration index than the Minnesota FDR section with a higher density (noted as HD in Figure 3.72), as expected. This shows that for similar material, the DCP is sensitive to changes in density under field testing conditions. Figure 3.73 shows that the DPI also reflects the influence of two other material properties. The NY 23A field project in New York showed that the DPI is sensitive to changes in curing time. (DPI decreased with increasing curing time, as expected.) In addition, the different support conditions from Indiana showed that the poor support section had a higher DPI value than the good support section, as might be expected if the underlying condition had an influence on the recycled material. (As shown in Table 2.3, the densities were the same.) 2 2 3 2 5 3 0 2 4 6 8 10 12 1 hour 1 hour 3 hours 3 hours 1 hour 1 hour SC 123 NM US 491 MN Cell 1-HD MN Cell 1-LD MN Cell 7 MN Cell 5 FDR F-C FDR F-C FDR E-C FDR E-C CCPR F-C CCPR E-N D PI , m m /b lo w Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-C = foam plus cement; E-C = emulsion plus cement; E-N = emulsion, no cement. Figure 3.72. DCP field testing results, FDR and CCPR mixtures. 1 1 3 1 2 1 3 1 2 2 0 2 4 6 8 10 12 1 hour 1.5 hours 1 hour 2 hours 1 hour 1 hour 1 hour 1 hour 1 hour 1 hour NY 23A NY 23A MN Cell 3 CA SR 22 NY 30 NY 28 MN Cell 4 IN SR 1- GS IN SR 1- PS CA SR 178 CIR F-N CIR F-N CIR F-C CIR F-C CIR E-N CIR E-N CIR E-N CIR E-N CIR E-N CIR E-C D PI , m m /b lo w Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-N = foam, no cement; F-C = foam plus cement; E-N = emulsion, no cement; E-C = emulsion plus cement. Figure 3.73. DCP field testing results, CIR mixtures.

88 Figure 3.73 shows that the DCP penetration index values for CIR projects were similar to or slightly higher than those for the FDR and CCPR projects. This is especially true for the SR 22 project from California that had the highest penetra- tion index of all projects. Figure 3.73 also shows the influence of two other material properties. The NY 23A field project from New York showed that the DCP penetration index was sensitive to changes in curing time, where the penetration index decreased with respect to curing time, as expected. In addition, the different support conditions from Indiana showed that the poor support section had a higher penetra- tion index value than the good support section, as might be expected if the underlying condition had an influence on the recycled material. (As shown in Table 2.3, the densities were the same.) 3.5.3 Shear Resistance 3.5.3.1a Number of Blows Figure 3.74 shows the number of blows from the LPST results for FDR and CCPR mixtures, and Figure 3.75 shows them for CIR mixtures. The figures show a similar range of results when FDR and CCPR are compared with CIR mix- tures. In addition, the lower-density section from Minnesota had fewer blows than the corresponding higher-density section, as expected. Figure 3.75 also shows a relatively wider range of test results for the CIR mixtures. Results from the NY 23A project indicated that the LPST number of blows was sensitive to changes in curing time in the field. The two support conditions from the Indiana project showed a similar number of blows. As with the DCP test results, the SR 22 project from California had the fewest blows as compared to the rest of the CIR projects. 3.5.3.2 Torque Value Figure 3.76 shows the LPST torque value results for FDR and CCPR mixtures, and Figure 3.77 shows them for CIR mixtures. The figures show a similar range of torque values for all three recycling processes. The lower-density FDR section from the Minnesota project had a lower torque value than the higher-density section, as expected. The two tests from the NY 23A project showed that the LPST torque value was sensitive to changes in curing. A similar LPST torque value was observed for the good and poor support conditions from the Indiana project. 3.5.4 Raveling Resistance 3.5.4.1a Number of Blows Figure 3.78 shows the SPRT number of blows for FDR and CCPR mixtures, and Figure 3.79 shows them for CIR mixtures. The number of blows shown from the field testing is the same as the number of blows (N2) shown from the laboratory testing. From Figure 3.78, the SPRT number of blows showed a ranking of projects similar to that of the LPST number of blows. As expected, the lower-density FDR section from the Minnesota project had fewer blows than the higher-density FDR section. 2 2 3 3 5 3 0 5 10 15 20 25 30 35 40 45 50 1 hour 1 hour 3 hours 3 hours 1 hour 1 hour SC 123 NM US 491 MN Cell 1-HD MN Cell 1-LD MN Cell 7 MN Cell 5 FDR F-C FDR F-C FDR E-C FDR E-C CCPR F-C CCPR E-N LP ST N um be r of B lo w s Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-C = foam plus cement; E-C = emulsion plus cement; E-N = emulsion, no cement. Figure 3.74. Long-pin shear test number of blows field testing results, FDR and CCPR mixtures.

89 1 1 3 1 2 1 3 1 2 2 0 5 10 15 20 25 30 35 40 45 50 1 hour 1.5 hours 1 hour 2 hours 1 hour 1 hour 1 hour 1 hour 1 hour 1 hour NY 23A NY 23A MN Cell 3 CA SR 22 NY 30 NY 28 MN Cell 4 IN SR 1- GS IN SR 1- PS CA SR 178 CIR F-N CIR F-N CIR F-C CIR F-C CIR E-N CIR E-N CIR E-N CIR E-N CIR E-N CIR E-C LP ST N um be r of B lo w s Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-N = foam, no cement; F-C = foam plus cement; E-N = emulsion, no cement; E-C = emulsion plus cement. Figure 3.75. Long-pin shear test number of blows field testing results, CIR mixtures. 2 2 3 3 5 3 0 20 40 60 80 100 120 140 160 180 1 hour 1 hour 3 hours 3 hours 1 hour 1 hour SC 123 NM US 491 MN Cell 1-HD MN Cell 1-LD MN Cell 7 MN Cell 5 FDR F-C FDR F-C FDR E-C FDR E-C CCPR F-C CCPR E-N LP ST T or qu e, ft -lb Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-C = foam plus cement; E-C = emulsion plus cement; E-N = emulsion, no cement. Figure 3.76. Long-pin shear test torque value field testing results, FDR and CCPR mixtures.

90 1 1 3 1 2 1 3 1 2 2 0 20 40 60 80 100 120 140 160 180 1 hour 1.5 hours 1 hour 2 hours 1 hour 1 hour 1 hour 1 hour 1 hour 1 hour NY 23A NY 23A MN Cell 3 CA SR 22 NY 30 NY 28 MN Cell 4 IN SR 1- GS IN SR 1- PS CA SR 178 CIR F-N CIR F-N CIR F-C CIR F-C CIR E-N CIR E-N CIR E-N CIR E-N CIR E-N CIR E-C LP ST T or qu e, ft -lb Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-N = foam, no cement; F-C = foam plus cement; E-N = emulsion, no cement; E-C = emulsion plus cement. Figure 3.77. Long-pin shear test torque value field testing results, CIR mixtures. 2 2 3 3 5 3 0 2 4 6 8 10 12 14 16 18 20 1 hour 1 hour 3 hours 3 hours 1 hour 1 hour SC 123 NM US 491 MN Cell 1-HD MN Cell 1-LD MN Cell 7 MN Cell 5 FDR F-C FDR F-C FDR E-C FDR E-C CCPR F-C CCPR E-N SP R T N um be r of B lo w s Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-C = foam plus cement; E-C = emulsion plus cement; E-N = emulsion, no cement. Figure 3.78. Short-pin raveling test number of blows field testing results, FDR and CCPR mixtures.

91 1 1 3 1 2 1 3 1 2 2 0 2 4 6 8 10 12 14 16 18 20 1 hour 1.5 hours 1 hour 2 hours 1 hour 1 hour 1 hour 1 hour 1 hour 1 hour NY 23A NY 23A MN Cell 3 CA SR 22 NY 30 NY 28 MN Cell 4 IN SR 1- GS IN SR 1- PS CA SR 178 CIR F-N CIR F-N CIR F-C CIR F-C CIR E-N CIR E-N CIR E-N CIR E-N CIR E-N CIR E-C SP R T N um be r of B lo w s Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-N = foam, no cement; F-C = foam plus cement; E-N = emulsion, no cement; E-C = emulsion plus cement. Figure 3.79. Short-pin raveling test number of blows field testing results, CIR mixtures. Figure 3.79 also shows that the projects with the greatest and fewest number of blows from the SPRT fixture were like those identified using the LPST fixture. The number of blows from the SPRT increased with respect to curing time, as seen from the NY 23A project. In addition, there is little difference in the number of blows from the Indiana CIR project for the good and poor support conditions. 3.5.4.2 Torque Value Figure 3.80 shows the SPRT torque values for FDR and CCPR mixtures, and Figure 3.81 shows them for CIR mix- tures. From Figure 3.80, the SRPT torque values show a relatively wider range of responses and a similar ranking of projects compared to the LPST. Figure 3.80 also shows some 2 2 3 3 5 3 0 5 10 15 20 25 30 35 40 45 50 1 hour 1 hour 3 hours 3 hours 1 hour 1 hour SC 123 NM US 491 MN Cell 1-HD MN Cell 1-LD MN Cell 7 MN Cell 5 FDR F-C FDR F-C FDR E-C FDR E-C CCPR F-C CCPR E-N SP R T To rq ue , f t-l b Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-C = foam plus cement; E-C = emulsion plus cement; E-N = emulsion, no cement. Figure 3.80. Short-pin raveling test torque value field testing results, FDR and CCPR mixtures.

92 differences between the high- and low-density FDR sections from Minnesota, but the differences are not likely to be statis- tically significant. Figure 3.81 shows that the SPRT torque values are also like the LPST torque values in terms of the rankings of the three projects with highest torque values. The LPST torque values are similar for seven of the 10 projects shown, but the SPRT torque values have a wider range over these same seven projects. The SPRT torque values also show a difference with respect to curing time for the NY 23A project and a slight difference in the torque values for the high- and low-density FDR sections from Minnesota. 3.5.5 Correlation Analysis A correlation analysis was performed to investigate the relationship between the tests performed in the Phase III field study. The analysis was performed by calculating the Pearson correlation coefficient (r) and the associated p-value. Table 3.59 shows the Pearson correlation coefficient and p-value for comparisons of the field-measured density, SSG stiffness, LWD modulus, LPST number of blows, LPST torque value, SPRT number of blows (N1 and N2), SPRT torque value, and DPI values. For those combinations that were shown to have a strong correlation (|r|>0.7,), the p-value was determined to estimate the significance of the relationship (alpha = 0.05). Shaded cells indicate comparisons where both conditions were met. The analysis showed that the following combinations had a strong, statistically significant correlation: • SSG stiffness with LWD modulus; • LPST number of blows with LPST torque value, SPRT torque value, and DPI; • LPST torque value with SPRT number of blows, SPRT torque value, and DPI; and • SPRT number of blows with SPRT torque value and DPI. Figures 3.82 through 3.90 demonstrate the relationship between those tests shown in the correlation analysis to have the strongest correlation and a statistically significant relation- ship based on the field testing. The data are presented along with a linear trendline to be consistent with the linear relation- ship shown by the Pearson correlation coefficient. For most comparisons, a linear trendline proved to have the highest coefficient of determination. However, for those comparisons including DPI, a nonlinear trend may prove to describe the relationships better. A small cluster of data artificially increased the correlation for certain comparisons (especially related to the LPST and SPRT results for NM US 491, NY SR 23A, and MN Cell 3). If the data for these three projects are removed from the analysis, only the SPRT torque values and number of blows were well correlated (i.e., |r|>0.7). Including all data shows that the LPST blows and torque, the SPRT blows and torque, and DPI were all well correlated. When the data from the three projects were removed, the correlation between the tests was reduced. 3.5.6 Lessons Learned During Field Testing Prior to conducting any of the tests in the field, suitable and uniform sites were selected based on visual observation of the recycling process and the completed recycled layer. As an example, cement as an active filler was observed to be applied non-uniformly across the width of the lane on 1 1 3 1 2 1 3 1 2 2 0 5 10 15 20 25 30 35 40 45 50 1 hour 1.5 hours 1 hour 2 hours 1 hour 1 hour 1 hour 1 hour 1 hour 1 hour NY 23A NY 23A MN Cell 3 CA SR 22 NY 30 NY 28 MN Cell 4 IN SR 1- GS IN SR 1- PS CA SR 178 CIR F-N CIR F-N CIR F-C CIR F-C CIR E-N CIR E-N CIR E-N CIR E-N CIR E-N CIR E-C SP R T To rq ue , f t-l b Curing Time; Project; Recycling Process, Agent, and Active Filler Notes: F-N = foam, no cement; F-C = foam plus cement; E-N = emulsion, no cement; E-C = emulsion plus cement. Figure 3.81. Short-pin raveling test torque value field testing results, CIR mixtures.

93 (a) SSG stiffness, MN/m LWD Modulus, ksi LPST Number of Blows LPST Torque, ft-lb SPRT Number of Blows SPRT Torque, ft-lb DPI, mm/blow Density, lb/ft3 0.0338 0.0831 0.5562 0.4963 0.5639 0.5389 −0.3611 SSG stiffness, MN/m 0.9106 0.6237 0.3298 0.5488 0.5134 −0.2496 LWD modulus, ksi 0.3686 0.0759 0.3189 0.2505 −0.0604 LPST number of blows 0.8839 −0.4291 0.8654 −0.7363 LPST torque, ft- lb 0.8863 0.8756 −0.7033 SPRT number of blows 0.9281 −0.7921 SPRT torque, ft- lb −0.6648 (b) SSG stiffness, MN/m LWD Modulus, ksi LPST Number of Blows LPST Torque, ft-lb SPRT Number of Blows SPRT Torque, ft-lb DPI, mm/blow Density, lb/ft3 0.9128 0.7596 0.0253 0.0505 0.0229 0.0313 0.1694 SSG stiffness, MN/m 0.0000 0.0227 0.2711 0.0521 0.0727 0.4108 LWD modulus, ksi 0.1601 0.7800 0.2287 0.3494 0.8243 LPST number of blows 0.0000 0.0972 0.0000 0.0011 LPST torque, ft-lb 0.0000 0.0000 0.0024 SPRT number of blows 0.0000 0.0003 SPRT torque, ft-lb 0.0050 Table 3.59. Field testing correlation analysis (a) Pearson correlation coefficient, (b) p-value. one field project. Thus, testing was conducted where the cement was observed to be applied uniformly. Also, if a large amount of loose material, cracks, segregated material, binder agglomerations, or crack sealant was observed, another location was selected. Testing was completed approximately within the center of the lane, and replicate tests were per- formed at a center-to-center spacing of approximately 1 ft to help ensure that testing was completed on the most uniform material. During stiffness testing with the SSG, surface preparation was important to obtaining test results having low variability. The surface to be tested was prepared by applying a thin layer of moist sand using a hand trowel such that any irre- gularities in the surface were filled with sand. The SSG foot was placed lightly on the sand patch, and the gauge was rotated approximately 90° clockwise, back to zero, then 90° counterclockwise, and then back to its original position with- out any downward force being applied to seat the foot. After each test, the foot was wiped clean with a rag and the sand patch was re-leveled with the hand trowel; more sand was added if needed. Stiffness testing using the LWD was influenced if the gauge was not solidly seated and if any hand pressure was applied to the LWD handle that resulted in a downward force. The LWD testing was conducted by placing the LWD on the surface and checking for a firm footing. If the LWD rocked back and forth, it was moved slightly until the rocking ceased. While the drops were being applied, the LWD was held still only by loosely circling the operator’s hand around the handle just below the top of the handle.

94 y = 1.2738x - 8.6304 R² = 0.8293 0 10 20 30 40 50 0 10 20 30 40 50 LW D M od ul us , k si SSG Stiffness, MN/m Figure 3.82. Relationship between soil stiffness gauge stiffness and LWD modulus. y = 2.4868x + 20.0703 R² = 0.7813 0 20 40 60 80 100 120 140 160 180 0 10 20 30 40 50 LP ST T or qu e, ft -lb LPST Number of Blows Figure 3.83. Relationship between long-pin shear test number of blows and long-pin shear test torque value. y = 0.7085x + 8.7120 R² = 0.7490 0 10 20 30 40 50 0 10 20 30 40 50 SP R T To rq ue , f t-l b LPST Number of Blows Figure 3.84. Relationship between long-pin shear test number of blows and short-pin raveling test torque value.

95 y = -0.1417x + 9.7401 R² = 0.5422 0 2 4 6 8 10 12 14 0 10 20 30 40 50 D PI , m m /b lo w LPST Number of Blows Figure 3.85. Relationship between long-pin shear test number of blows and DPI. y = 0.1150x - 0.1874 R² = 0.7855 0 5 10 15 20 0 20 40 60 80 100 120 140 160 180 SP R T N um be r of B lo w s LPST Torque, ft-lb Figure 3.86. Relationship between long-pin shear test torque value and short-pin raveling test number of blows. y = 0.2548x + 5.4441 R² = 0.7667 0 10 20 30 40 50 0 20 40 60 80 100 120 140 160 180 SP R T To rq ue , f t-l b LPST Torque, ft-lb Figure 3.87. Relationship between long-pin shear test torque value and short-pin raveling test torque value.

96 y = -0.0481x + 10.1612 R² = 0.4946 0 2 4 6 8 10 12 14 0 20 40 60 80 100 120 140 160 180 D PI , m m /b lo w LPST Torque, ft-lb Figure 3.88. Relationship between long-pin shear test torque value and DPI. y = 2.0806x + 7.0908 R² = 0.8613 0 10 20 30 40 50 0 5 10 15 20 SP R T To rq ue , f t-l b SPRT Number of Blows Figure 3.89. Relationship between short-pin raveling test number of blows and short-pin raveling test torque value. y = -0.4173x + 10.0762 R² = 0.6274 0 2 4 6 8 10 12 14 0 5 10 15 20 D PI , m m /b lo w SPRT Number of Blows Figure 3.90. Relationship between short-pin raveling test number of blows and DPI.

97 Penetration resistance testing using the DCP was com- pleted by ensuring that the DCP handle was kept plumb and that testing was not started on a large piece of aggregate at the pavement surface. DCP testing could be conducted by two operators, but a third was helpful: one holding the handle, one operating the weight, and a third reading the penetration depth. Recent commercial developments to the DCP include automated counters, which could reduce the number of operators needed. Operating the DCP by recording the dis- tance to a predetermined number of blows was considered as an option but ultimately was not selected for reasons discussed Section 3.6.1: Proposed Tests. Shear and raveling resistance tests were completed by including a 1-in.-diameter bubble level to ensure that the base plate was kept parallel to the surface of the recycled layer while the shear and raveling fixtures were driven into the recycled material. For both tests, the operator listened for a change in the sound while driving the fixture to note when the base plate was touching the surface of the recycled layer. The rate at which the torque was applied was kept constant from location to location by drawing a line (using a lumber crayon on the recycled surface) 12 in. long from the center of the base plate. Perpendicular to this 12-in. line, and at the end farthest from the center of the base plate, a 4-in. line was drawn in the direction the torque wrench was to be pulled by the operator. During testing, the operator counted 4 seconds and, during this time, swung the arm of the torque wrench along the entire length of the 4-in. line. This method allowed the operator to apply a consistent torque rate, and the maximum value was always obtained within this distance. Figure 3.91 shows an illustration of this process. During initial stages of the field testing, there was concern that the recycled material might be susceptible to raveling after completion of the field tests given that most of the projects were opened to traffic soon after the tests were com- pleted. The most destructive test devices included the DCP and the shear and raveling fixtures. Following each of these tests, any disturbed material could be tamped back into place by the operator simply stepping on and compressing the disturbed material. Figure 3.92 shows the typical condition of the pavement surface just after testing and again after the disturbed material had been tamped. Figure 3.93 shows the condition of the pavement surface 4 days after testing having no deterioration caused by the testing. 4 inches Shear or raveling fixture 12 inc hes Figure 3.91. Plan view of torque application. Figure 3.92. Recycled material after a torque test (left) and after tamping (right).

98 3.5.7 Preliminary Precision Statements as Determined During Interlaboratory Study As part of the field testing, the research team conducted an ILS to develop preliminary precision statements for the shear and raveling tests developed in this study. The term “preliminary” is used since only three laboratories participated in the ILS and the ILS was conducted in the field. The ILS was conducted to develop preliminary precision statements for penetration resistance testing using the DCP, number of blows and torque value for the LPST, and number of blows and torque value for the SPRT. Following a presentation of the data for each test, details of determining the data consistency, the form of the precision statement, and the final precision statements are provided. 3.5.7.1 Preliminary Precision Statement for DCP Testing The average DPI value from the ILS testing by all participat- ing laboratories and test cells is shown in Table 3.60. Testing using the DCP was conducted in only four of six test cells. Prior to the collected data being analyzed, the single- operator and between-laboratory consistency were investi- gated with respect to the average and dispersion of the results. This check was performed to prevent effects of any potential inconsistent data on the precision of the test method. Data consistency was checked in accordance with the procedure outlined in ASTM C802. Two statistical parameters are defined in ASTM C802 to evaluate data consistency: the k-value and the h-value. The k-value is used to check the consistency of the single- operator variability for each laboratory for a given material. The k-values are always positive numbers and indicate how the variability of a laboratory might be different from the variability of other laboratories in a study (i.e., pooled vari- ability). High k-values denote high single-operator variability. The h-value is used to investigate whether the average value of a laboratory is consistent with the overall average of the other laboratories for a given material. Unlike the k-values, the h-values can be positive or negative numbers. Positive or negative h-values show that an average property measure of a laboratory is larger or smaller than the average property measures of other laboratories. In addition, outliers are determined on the basis of checking the variability values (k-value and h-value) of a laboratory (or material) with respect to critical values (k-value and h-value) determined at a given significance level (e.g., 0.5%). These calculated k- and h-values were checked with respect to the critical k- and h-values determined on the basis of the Figure 3.93. Test area four days after a torque test (left) and close-up of the same location (right). Laboratory Replicates DPI, mm/blow Cell 1 FDR E-C Cell 3 CIR F-C Cell 4 CIR E-N Cell 7 CCPR F-C Lab 1 1 5.3 4.6 5.1 5.5 2 4.8 4.4 5.0 5.2 3 4.6 4.1 5.5 5.8 Lab 2 1 5.1 3.8 6.2 4.9 2 4.4 5.0 7.5 5.3 3 5.3 4.0 8.7 5.2 Lab 3 1 4.7 4.5 6.2 6.1 2 4.7 4.8 5.7 6.0 3 4.7 4.2 5.6 6.0 Table 3.60. Summary of DPI from ILS.

99 number of replicates used and the number of laboratories participating in the ILS. ASTM C802 provides a table of criti- cal values of k and h statistics at a 0.5% significance level. The critical k-values are a function of the number of participating laboratories and the number of replicate test measurements. On the other hand, the critical h-values depend only on the number of participating laboratories. Throughout this study, there were three participating laboratories for any of the tests considered, with three replicate measurements for each test unless otherwise indicated. Hence, the critical k-value for this ILS was determined to be 1.67, whereas the critical h-value equals ± 1.15. Both values were determined in accordance with ASTM C802. Tables 3.61 and 3.62 show the analysis of data consistency for DPI in terms of the calculated k-values and h-values, respectively. The results show that all laboratory/material combinations were within acceptable limits when compared to the critical values and therefore indicate consistency in the collected data. The data were also investigated to identify if any inter- actions among laboratories and test cells existed. The presence of interactions was investigated by determining whether the pattern of the test results obtained on the section by one laboratory differed notably from the pattern obtained by the other laboratories. Figure 3.94 shows the DPI values for each laboratory and material combination with respect to the test cell from lowest to highest values. The average DPI values and the value from each laboratory with respect to the test cell followed a similar trend. However, the DPI value for Labora- tory 2 and Cell 4 was significantly higher than for the other laboratories. Although it is unclear as to why such a trend was observed, the data were included in the analysis as the consistency statistics (k- and h-values) of single-operator and between-laboratory conditions were within the acceptable limits, as presented previously, and the number of participating laboratories was already limited. The single-operator and multi-laboratory standard devia- tion and COV for DPI with respect to test cells are presented in Table 3.63, as calculated in accordance with ASTM C802. To determine the form of the precision statements, the relationship between the average DPI values and the stan- dard deviation and the COV for single-operator and multi- laboratory conditions was investigated and is shown in Figures 3.95 and 3.96, respectively. The single-operator standard deviation and COV increased slightly with increasing DPI values. The multi-laboratory standard deviation and COV also increased with increasing DPI values. However, there were only four observations, and the increasing trend for the multi-laboratory condition is highly affected by the observation from the Laboratory 2 Laboratory DPI, mm/blow Cell 1 FDR E-C Cell 3 CIR F-C Cell 4 CIR E-N Cell 7 CCPR F-C Lab 1 1.07 0.64 0.35 1.42 Lab 2 1.36 1.46 1.65 0.97 Lab 3 0.07 0.69 0.38 0.23 Note: Critical k-value equals 1.67. Table 3.61. k-values for single-operator data consistency for DPI. Laboratory DPI, mm/blow Cell 1 FDR E-C Cell 3 CIR F-C Cell 4 CIR E-N Cell 7 CCPR F-C Lab 1 0.50 0.05 −0.83 −0.11 Lab 2 0.65 −1.02 1.11 −0.94 Lab 3 −1.15 0.98 −0.29 1.05 Note: Critical h-value equals ± 1.15. Table 3.62. h-values for between-laboratory data consistency for DPI. 0 1 2 3 4 5 6 7 8 D PI , m m /b lo w Test Cell Lab1 Lab2 Lab3 Cell 3 Cell 1 Cell 7 Cell 4 Figure 3.94. Average DPI values arranged from least to greatest.

100 R² = 0.27 R² = 0.67 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 St an da rd D ev ia tio n, m m /b lo w DPI, mm/blow Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) 3.5 4.0 4.5 5.0 5.5 6.0 6.5 Figure 3.95. Relationship between average DPI measurements and standard deviation. R² = 0.04 R² = 0.56 0 5 10 15 20 25 C oe ff ic ie nt o f V ar ia tio n, % DPI, mm/blow Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) 3.5 4.0 4.5 5.0 5.5 6.0 6.5 Figure 3.96. Relationship between average DPI measurements and coefficient of variation. Test Cell/Material DPI, mm/blow Average Standard Deviation Coefficient of Variation, % Single Operator Multi- Laboratory Single Operator Multi- Laboratory Cell 3 CIR F-C 4.4 0.43 0.43 9.8 9.8 Cell 1 FDR E-C 4.8 0.34 0.34 7.0 7.0 Cell 7 CCPR F-C 5.5 0.22 0.47 4.0 8.6 Cell 4 CIR E-N 6.2 0.76 1.33 12.3 21.6 Table 3.63. DPI average, standard deviation, and coefficient of variation.

101 measurement on Cell 4. Given this slight increase and effect of the one observation in Cell 4, the assumption of constant standard deviation or COV was made. Since the COV tends to be more independent than the standard deviation, the use of a constant COV is appropriate for developing the precision statements for DCP test results. The pooled single-operator and multi-laboratory COV values from Table 3.63 were used to develop the precision statements for DCP measurements. The precision statements for DCP measurements were devel- oped in accordance with ASTM C670 and are presented here. • Single-operator precision: The single-operator COV was 8.3%. Therefore, the results of two properly conducted tests by the same operator on the same material are not expected to differ by more than 23.2%A of their average. • Multi-laboratory precision: The multi-laboratory COV was 11.7%. Therefore, the results of two properly con- ducted tests by two different laboratories on specimens of the same material are not expected to differ by more than 32.8%A of their average. AThese numbers represent the difference limits in % (d2s%) as described in ASTM Practice C670. Note: These precision statements are based on an ILS that involved three laboratories, four materials, and three replicate tests per operator, with DPIs ranging from 3.8 to 8.7 mm/blow. 3.5.7.2 Preliminary Precision Statement for Shear Tests The average number of blows and torque value from testing from the LPST are shown in Tables 3.64 and 3.65, respectively. Summaries of calculated k-values and h-values for the LPST number of blows for each test cell are shown in Tables 3.66 and 3.67, respectively. The values that exceeded the critical values are highlighted and bolded in the tables. The only laboratory/material combination that exceeded a critical statistic parameter (k-value in this case of more than 1.67) was Laboratory 1 for Cell 1 (FDR E-C LD). A parameter exceeding the critical value normally calls for elimination of the data from the remaining part of the analysis. However, the high variability for this section was attributed to likely nonhomogenous characteristics of the section. This partic- ular test area was an unintended construction anomaly iden- tified by the research team within a planned test section, and thus higher variability might be expected. Given the limited number of laboratories and sections, the research team opted to include these data for further analysis. It was expected that the resulting precision values would not be adversely affected by the inclusion of these data in the analysis. Similarly, Tables 3.68 and 3.69 show the calculated k-values and h-values for the LPST torque values for each test cell, respectively. All the h- values and k-values were within the acceptable range except the k-value for Cell 1 (FDR E-C LD) Laboratory Replicates Number of Blows Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 1 17 9 44 20 21 30 2 16 12 40 19 22 30 3 18 16 40 20 22 30 Lab 2 1 17 10 42 18 21 29 2 17 11 37 18 23 28 3 17 11 43 18 22 28 Lab 3 1 18 17 47 19 28 36 2 24 16 47 19 24 34 3 20 16 43 18 24 32 Table 3.64. Summary of LPST number of blows from ILS. Laboratory Replicates Torque, ft-lb Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 1 68.9 47.3 133.3 76.9 54.5 91.9 2 89.9 39.5 161.1 78.0 52.2 93.7 3 83.1 73.3 157.7 66.3 62.3 92.0 Lab 2 1 72.1 49.1 154.4 66.5 54.8 77.0 2 82.0 51.1 162.5 71.8 55.6 83.0 3 80.7 52.4 165.6 64.9 63.4 84.1 Lab 3 1 92.4 75.0 133.4 76.7 100.2 113.0 2 101.0 76.8 149.4 81.8 71.7 108.2 3 78.0 75.1 139.8 79.3 70.9 106.8 Table 3.65. Summary of LPST torque values from ILS.

102 by Laboratory 1, with a k-value of 1.69, the same observation as with the case of number of blows measurements. For the same reasons as stated previously, the data associated with this observation were not removed from the analysis. To evaluate interactions among laboratories and test cells, the LPST number of blows and torque value are shown in Figures 3.97 and 3.98, respectively. The data are arranged from least to greatest value. The trends from each laboratory were similar for both the number of blows and torque value. From the results shown in Figures 3.97 and 3.98, no data were excluded because of interactions. Tables 3.70 and 3.71 show the single-operator and multi- laboratory standard deviation and COV for LPST number of blows and torque value, respectively, as calculated for each test cell in accordance with ASTM C802. To determine the form of the precision statements, the relationship between the average LPST number of blows and the standard deviation and the COV for single-operator and multi-laboratory conditions is shown in Figures 3.99 and 3.100, respectively. The figures indicate that the stan- dard deviation is also the appropriate basis for developing the precision statements for LPST number of blows mea- surements. That is because the standard deviation tends to be relatively more independent of LPST number of blows measurements than the COV for both single-operator and multi-laboratory conditions. Thus, the pooled single-operator Laboratory Number of Blows Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 0.54 1.69 0.87 1.22 0.39 0.00 Lab 2 0.00 0.28 1.21 0.00 0.67 0.48 Lab 3 1.65 0.28 0.87 1.22 1.55 1.66 Notes: Critical k-value equals 1.67; bold/highlight = critical value exceeded. Table 3.66. k-values for single-operator data consistency for LPST number of blows. Laboratory Number of Blows Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 −0.58 −0.27 −0.45 1.06 −0.66 −0.27 Lab 2 −0.58 −0.84 −0.70 −0.93 −0.49 −0.84 Lab 3 1.15 1.11 1.15 −0.13 1.15 1.11 Note: Critical h-value equals ± 1.15. Table 3.67. h-values for between-laboratory data consistency for LPST number of blows. Laboratory Torque Value, ft-lb Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 1.11 1.72 1.45 1.39 0.50 0.34 Lab 2 0.56 0.16 0.55 0.78 0.45 1.29 Lab 3 1.21 0.10 0.77 0.67 1.60 1.10 Notes: Critical k-value equals 1.67; bold/highlight = critical value exceeded. Table 3.68. k-Values for single-operator data consistency for LPST torque values. Laboratory Torque Value, ft-lb Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 −0.38 −0.48 −0.01 0.05 −0.64 −0.13 Lab 2 −0.75 −0.67 1.00 −1.02 −0.52 −0.93 Lab 3 1.14 1.15 −0.99 0.97 1.15 1.06 Note: Critical h-value equals ± 1.15. Table 3.69. h-values for between-laboratory data consistency for LPST torque values.

103 0 5 10 15 20 25 30 35 40 45 50 Cell 1LD Cell 1 Cell 4 Cell 5 Cell 7 Cell 3 LP ST N um be r of B lo w s Test Cell Lab1 Lab2 Lab3 Figure 3.97. Average LPST number of blows arranged from least to greatest. 0 20 40 60 80 100 120 140 160 180 Cell 1LD Cell 5 Cell 4 Cell 1 Cell 7 Cell 3 LP ST T or qu e, ft -lb Test Cell Lab 1 Lab 2 Lab 3 Figure 3.98. Average LPST torque arranged from least to greatest. Test Cell/Material Number of Blows Average Standard Deviation Coefficient of Variation, % Single Operator Multi- Laboratory Single Operator Multi- Laboratory Cell 1 FDR E-C LD 13.1 2.1 3.4 15.9 25.7 Cell 1 FDR E-C 18.2 1.9 2.6 10.2 14.3 Cell 4 CIR E-N 18.8 0.5 0.9 2.5 4.9 Cell 5 CCPR E-N 23.0 1.5 2.4 6.5 10.3 Cell 7 CCPR F-C 30.8 1.2 3.1 3.9 10.0 Cell 3 CIR F-C 42.6 2.6 3.5 6.2 8.2 Table 3.70. LPST number of blows average, standard deviation, and coefficient of variation.

104 Test Cell/Material Torque Value, ft-lb Average Standard Deviation Coefficient of Variation, % Single Operator Multi- Laboratory Single Operator Multi- Laboratory Cell 1 FDR E-C LD 60.0 10.3 16.0 17.2 26.7 Cell 5 CCPR E-N 65.1 10.6 16.2 16.2 24.8 Cell 4 CIR E-N 73.6 4.6 6.8 6.3 9.2 Cell 1 FDR E-C 83.1 9.6 10.2 11.6 12.3 Cell 7 CCPR F-C 94.4 3.0 14.3 3.1 15.1 Cell 3 CIR F-C 150.8 10.5 13.1 6.9 8.7 Table 3.71. LPST torque value average, standard deviation, and coefficient of variation. R² = 0.16 R² = 0.16 0 1 2 3 4 5 6 0 10 20 30 40 50 St an da rd D ev ia tio n LPST Number of Blows Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.99. Relationship between average LPST number of blows and standard deviation. R² = 0.24 R² = 0.29 0 5 10 15 20 25 30 0 10 20 30 40 50 C oe ff ic ie nt o f V ar ia tio n, % LPST Number of Blows Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.100. Relationship between average LPST number of blows and coefficient of variation.

105 and multi-laboratory standard deviations from Table 3.70 were used to develop the precision statements for LPST number of blows measurements. Similarly, the relationship between the average LPST torque value and corresponding standard deviation for single-laboratory and multi-laboratory conditions is shown in Figure 3.101. The relationship between average LPST number of blows and COV for single-laboratory and multi- laboratory conditions is shown in Figure 3.102. From these two figures, it is evident that the standard deviation was independent of the LPST torque measurements for both single-operator and multi-laboratory conditions, whereas (although not a strong relationship) the single-operator and multi-laboratory COV tended to decrease with an increase in LPST torque measurements. Hence, the use of a constant standard deviation is appropriate for developing the precision statements for LPST torque results. Therefore, the pooled single-operator and multi-laboratory standard deviations from Table 3.71 were used to develop the precision statements for LPST torque values. The following precision statements for LPST number of blows and torque value were developed in accordance with ASTM C670. Number of Blows • Single-operator precision: The single-operator standard deviation was 1.6 blows. Therefore, the results of two properly conducted tests by the same operator on the same material are not expected to differ by more than five blows.A R² = 0.01 R² = 0.004 0 2 4 6 8 10 12 14 16 18 40 60 80 100 120 140 160 180 200 220 St an da rd D ev ia tio n LPST Torque, ft-lb Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.101. Relationship between average LPST torque value and standard deviation. R² = 0.32 R² = 0.42 0 5 10 15 20 25 30 0 20 40 60 80 100 120 140 160 180 C oe ff ic ie nt o f V ar ia tio n, % LPST Torque, ft-lb Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.102. Relationship between average LPST torque value and coefficient of variation.

106 • Multi-laboratory precision: The multi-laboratory stan- dard deviation was 2.6 blows. Therefore, the results of two properly conducted tests by two different laboratories on specimens of the same material are not expected to differ by more than seven blows.A AThese numbers represent the difference limits (d2s) as described in ASTM C670. Note: These precision statements are based on an ILS that involved three laboratories, six materials, and three replicate tests per operator, with number of blows values ranging from nine to 54. Torque Value • Single-operator precision: The single-operator standard deviation was 8.1 ft-lbf. Therefore, the results of two properly conducted tests by the same operator on the same material are not expected to differ by more than 22.6 ft-lbf.A • Multi-laboratory precision: The multi-laboratory stan- dard deviation was 12.7 ft-lbf. Therefore, the results of two properly conducted tests by two different laboratories on specimens of the same material are not expected to differ by more than 35.7 ft-lbf.A AThese numbers represent the difference limits (d2s) as described in ASTM C670. Note: These precision statements are based on an ILS that involved three laboratories, six materials, and three replicate tests per operator, with torque values ranging from 39.5 ft-lbf to 165.6 ft-lbf. 3.5.7.3 Preliminary Precision Statement for Raveling Tests Tables 3.72 and 3.73 show the SPRT number of blows and torque value, respectively. Summaries of calculated k-values and h-values for the SPRT number of blows for each test cell are shown in Tables 3.74 and 3.75, respectively. The values that exceeded the critical values are highlighted and bolded in the tables. For reasons stated previously, the data associated with this observation were included in the analysis. Similarly, Tables 3.76 and 3.77 show the calculated k-values and h-values for the SPRT torque values for each test cell, respectively. As shown, none of the k-values and h-values exceeded the critical values, indicating consistency in the collected data. To evaluate interactions among laboratories and test cells, the SPRT number of blows and torque value are shown in Figures 3.103 and 3.104, respectively. The data are arranged from least to greatest value. As seen from the figures, the trends from each laboratory were similar for both the number of blows and torque value. No data were excluded because of interactions. Laboratory Replicates Material Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 1 8 7 15 8 9 11 2 7 6 19 8 9 13 3 10 7 16 8 9 14 Lab 2 1 6 4 16 8 7 9 2 6 4 15 7 8 10 3 7 4 15 7 8 10 Lab 3 1 8 6 16 7 8 11 2 8 6 15 8 9 10 3 8 6 14 7 9 10 Table 3.72. Summary of SPRT number of blows from ILS. Laboratory Replicates Torque Value, ft-lb Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 1 26.1 22.9 39.4 25.6 20.7 27.2 2 26.2 17.8 46.3 22.5 23.3 24.0 3 23.9 15.2 42.0 18.8 21.0 28.8 Lab 2 1 18.5 18.1 50.6 29.7 21.9 26.1 2 19.7 15.9 44.5 21.7 16.6 23.2 3 20.6 16.6 40.9 26.5 20.0 23.8 Lab 3 1 22.8 21.8 50.8 24.8 25.9 30.2 2 22.9 27.4 45.7 26.8 18.9 28.6 3 19.8 21.8 40.0 24.8 19.6 25.3 Table 3.73. Summary of SPRT torque value from ILS.

107 Laboratory Number of Blows Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 1.62 1.73 1.51 0.00 0.00 1.53 Lab 2 0.61 0.00 0.42 1.22 1.22 0.58 Lab 3 0.00 0.00 0.73 1.22 1.22 0.58 Notes: Critical k-value equals 1.67; bold/highlight = critical value exceeded. Table 3.74. k-values for single-operator data consistency for SPRT number of blows. Laboratory Number of Blows Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 0.73 0.80 1.13 1.15 0.80 1.13 Lab 2 −1.14 −1.12 −0.38 −0.58 −1.12 −0.78 Lab 3 0.41 0.32 −0.76 −0.58 0.32 −0.35 Note: Critical h-value equals ± 1.15. Table 3.75. h-values for between-laboratory data consistency for SPRT number of blows. Laboratory Torque Value, ft-lb Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 0.93 1.30 0.75 1.09 0.50 1.11 Lab 2 0.75 0.37 1.05 1.29 0.95 0.69 Lab 3 1.26 1.08 1.16 0.37 1.36 1.13 Note: Critical k-value equals 1.67. Table 3.76. k-values for single-operator data consistency for SPRT torque values. Laboratory Torque Value, ft-lb Cell 1 FDR E-C Cell 1 FDR E-C LD Cell 3 CIR F-C Cell 4 CIR E-N Cell 5 CCPR E-N Cell 7 CCPR F-C Lab 1 1.07 −0.31 −1.15 −1.15 0.66 0.17 Lab 2 −0.92 −0.81 0.53 0.70 −1.15 −1.07 Lab 3 −0.15 1.12 0.63 0.45 0.49 0.91 Note: Critical h-value equals ± 1.15. Table 3.77. h-values for between-laboratory data consistency for SPRT torque values. The single-operator and multi-laboratory standard devia- tion and COV for SPRT number of blows and torque value are presented in Tables 3.78 and 3.79, respectively, as calculated for each test cell in accordance with ASTM C802. To determine the form of the precision statements, the relationship between the average SPRT number of blows and the standard deviation and the COV for single-operator and multi-laboratory conditions is shown in Figures 3.105 and 3.106, respectively. The figures show that the COV is the appropriate basis for developing the precision statements for SPRT number of blows as the COV overall tended to be relatively more independent than the standard deviation for both single-operator and multi-laboratory conditions. Thus, the pooled single-operator and multi-laboratory standard deviations from Table 3.78 were used to develop the precision statements for SPRT number of blows. The relationship between average SPRT torque values and the standard deviation and the COV for single-operator and multi-laboratory conditions is presented in Figures 3.107 and 3.108, respectively. The standard deviation tended to increase with an increase in the SPRT torque value for both single-operator and multi-laboratory conditions, and the single-operator and multi-laboratory COV stayed relatively constant with changes in the LPST torque value. Thus, the use of a constant COV is appropriate for developing the precision statements for SPRT torque values. The pooled single-operator and multi-laboratory COVs from Table 3.79 were used to develop the precision statements for SPRT torque values.

108 0 2 4 6 8 10 12 14 16 18 Cell 1LD Cell 4 Cell 1 Cell 5 Cell 7 Cell 3 SP R T N um be r of B lo w s Lab1 Lab2 Lab3 Test Cell Figure 3.103. Average SPRT number of blows arranged from least to greatest. 0 5 10 15 20 25 30 35 40 45 50 Cell 1LD Cell 5 Cell 1 Cell 4 Cell 7 Cell 3 SP R T To rq ue , f t-l b Test Cell Lab1 Lab2 Lab3 Figure 3.104. Average SPRT torque value arranged from least to greatest. Test Cell/Material Number of Blows Average Standard Deviation Coefficient of Variation, % Single Operator Multi- Laboratory Single Operator Multi- Laboratory Cell 1 FDR E-C LD 5.6 0.3 1.4 6.0 25.5 Cell 4 CIR E-N 7.6 0.5 0.5 6.2 7.2 Cell 1 FDR E-C 7.6 0.9 3.3 12.5 43.4 Cell 5 CCPR E-N 8.4 0.5 0.8 5.6 9.4 Cell 7 CCPR F-C 10.9 1.0 1.8 9.2 16.3 Cell 3 CIR F-C 15.7 1.4 1.4 8.8 9.1 Table 3.78. SPRT number of blows average, standard deviation, and coefficient of variation.

109 Test Cell/Material Torque Value, ft-lb Average Standard Deviation Coefficient of Variation, % Single Operator Multi- Laboratory Single Operator Multi- Laboratory Cell 1 FDR E-C LD 19.7 3.0 4.3 15.2 21.8 Cell 5 CCPR E-N 20.9 2.8 2.8 13.6 13.6 Cell 1 FDR E-C 22.3 1.4 3.1 6.3 14.1 Cell 4 CIR E-N 24.6 3.1 3.2 12.7 13.1 Cell 7 CCPR F-C 26.4 2.2 2.6 8.4 9.8 Cell 3 CIR F-C 44.5 4.7 5.4 10.5 12.3 Table 3.79. SPRT torque value average, standard deviation, and coefficient of variation. R² = 0.77 R² = 0.00 0 1 2 3 4 4 6 8 10 12 14 16 18 St an da rd D ev ia tio n SPRT Torque, ft-lb Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.105. Relationship between average SPRT number of blows and standard deviation. R² = 0.06 R² = 0.18 0 3 6 9 12 15 18 21 24 27 4 6 8 10 12 14 16 18 C oe ff ic ie nt o f V ar ia tio n, % SPRT Number of Blows Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.106. Relationship between average SPRT number of blows and coefficient of variation.

110 The following precision statements for SPRT number of blows and torque value were developed in accordance with ASTM C670. Number of Blows • Single-operator precision: The single-operator COV was 8%. Therefore, the results of two properly conducted tests by the same operator on the same material are not expected to differ by more than 22.5%A of their average. • Multi-laboratory precision: The multi-laboratory COV was 14.2%. Therefore, the results of two properly conducted tests by two different laboratories on specimens of the same material are not expected to differ by more than 39.6%A of their average. AThese numbers represent the difference limits in % (d2s%) as described in ASTM C670. Note: These precision statements are based on an ILS that involved three laboratories, six materials and three replicate tests per operator, with number of blows ranging from four to 19. Torque Value • Single-operator precision: The single-operator COV was 11.1%. Therefore, the results of two properly conducted tests by the same operator on the same material are not expected to differ by more than 31.1%A of their average. • Multi-laboratory precision: The multi-laboratory COV was 13.8%. Therefore, the results of two properly conducted R² = 0.56 R² = 0.51 0 1 2 3 4 5 6 10 20 30 40 50 St an da rd D ev ia tio n SPRT Torque, ft-lb Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.107. Relationship between average SPRT torque value and standard deviation. R² = 0.05 R² = 0.18 0 3 6 9 12 15 18 21 24 27 10 20 30 40 50 C oe ff ic ie nt o f V ar ia tio n, % SPRT Torque, ft-lb Single-Operator Multi-Laboratory Linear (Single-Operator) Linear (Multi-Laboratory) Figure 3.108. Relationship between average SPRT torque value and coefficient of variation.

111 tests by two different laboratories on specimens of the same material are not expected to differ by more than 38.7%A of their average. AThese numbers represent the difference limits in % (d2s%) as described in ASTM C670. Note: These precision statements are based on an ILS that involved three laboratories, six materials, and three replicate tests per operator, with torque values ranging from 15.2 to 50.8 ft-lbf. 3.6 Selection of Recommended Tests and Threshold Values Selecting the recommended tests and their corresponding threshold values was conducted by comparing the different tests, subjectively considering the ease at which the tests could be conducted, and making a statistical evaluation of the test variability. Due to differences in the confinement conditions between laboratory testing (Phase II) and field testing (Phase III), the proposed test selection and threshold values were determined based on field test data. 3.6.1 Proposed Tests From the findings of this study, the SSG and LWD tests were not recommended since they were found to be influ- enced by the properties of the pavement foundation and not just the recycled layer. This was evidenced by the large difference in measured stiffness properties of the Indiana SR 1 project having good and poor underlying support layers. If the goal of this project were to include an overall structural capacity assessment that included the foundation, the SSG and LWD tests would have been good candidates. The DPI was found to be somewhat correlated with the remaining tests (LPST number of blows and torque value, SPRT number of blows and torque value) when considering the coefficient of determination (calculated by squaring the values shown in Table 3.59a). The coefficient of determina- tion (R2) when considering the DPI versus the remaining tests was found to range from 0.44 to 0.63, indicating that approxi- mately 44% to 63% of the variation in the other tests was explained by the DPI values. Because of this lower explanatory power, and because the DPI is only a measure of penetration resistance and not any type of shearing force, the DPI was not recommended. Because of their ability to measure independent material properties and the ease at which the test can be conducted, the LPST number of blows and torque and the SPRT number of blows and torque were recommended. Despite the good correlation shown during the Phase III field testing between the number of blows and torque value for each fixture, these two components of each test are recommended since the number of blows can easily be counted while preparing to assess the torque value. In addition, as shown in the next section, not all projects respond to the number of blows and torque threshold values the same way despite the good correlation. This added a level of conservatism to the testing. The LPST number of blows and torque are thought to be an indicator of the material’s ability to withstand the loading from traffic and from paving equipment when the recycled material is surfaced. The SPRT is thought to be an indicator that the recycled material can be trafficked without experi- encing deterioration of the surface of the recycled layer. 3.6.2 Proposed Threshold Values A statistical approach was used to establish the threshold values for each recommended test. The threshold value is that value that defines the separation between deficient material and adequate material. Threshold values were calculated using the mean and variability of each test from Phase III using an approach similar to the percent within limits (PWL) concept. The PWL is the proportion of test values that fall within a predetermined upper and lower specification limit. Hughes (1996) and Muench and Mahoney (2001) state that when agencies work to develop statistical-based quality assur- ance programs, these specification limits are set most often based on two factors: engineering judgment and statistical analysis. In an ideal case, the proportion considered acceptable should be set such that the maximum amount of accepted defective material will not substantially degrade the overall quality of the pavement. Typical levels of acceptable quality for pavement materials are often set such that 90% or 95% of test values from a given population are identified as adequate (Muench and Mahoney, 2001). Using these concepts, a lower specification limit was calculated for each test and was con- sidered as the threshold value. Using engineering judgment and concepts from Hughes (1996), the threshold values for each test were calculated by considering the left-side tail of the field dataset distribution as shown in Figure 3.109. The threshold value was calculated Mean Threshold value 5% Figure 3.109. Example normal distribution with threshold value at a left-tail area of 5%.

112 by first determining the mean value from all 50 observations of each test. Next, the standard deviation from each project (termed “project standard deviation”) was calculated based on their respective replicate measurements. Finally, the pooled standard deviation was calculated as the average of the project standard deviations. The threshold value was calculated such that 95% of the test observations would be found adequate (5% deficient) as follows: TV = Mean − 1.645(PSD) Where TV = threshold value, mean = mean value, 1.645 = z-statistic for a left-tail area of 5%, and PSD = pooled stan- dard deviation. Table 3.80 shows the mean, pooled standard deviation, and threshold values for each test. Given the rela- tively low number of projects for each recycling process and stabilizing/recycling agent combination, the same threshold values were suggested regardless of recycling process and stabilizing/recycling agent combination. To add a layer of conservatism, a material must pass both the number of blows and torque component to be considered as adequate. The threshold values shown in Table 3.80 were compared to the test results from two field projects that had the lowest values and from two additional trials where 100% RAP was placed but no additional binder was added. The two 0% binder trials, one conducted in the laboratory and one in the field, were used to simulate a worst-case scenario with respect to cold recycled materials—the case where no additional binder is added. Table 3.81 shows the test results from the Phase III field projects that, while constructed successfully, had the lowest measured values for three of the four recommended tests. The CA SR 22 project was tested at three curing times while the MN Cell 1-LD site was only tested at one curing time. Comparing the test results from these two projects to the proposed threshold values shows that the results of the four tests did not meet the threshold value when testing was conducted at the earliest curing time. This suggests that both surfacing and trafficking should wait to see if additional curing improves the test results. Only the CA project was assessed at longer curing intervals and Table 3.81 also shows that the test results met or exceeded all threshold values after 24 hours. Table 3.82 shows the test results from the two 0% binder trials. For the laboratory assessment, RAP from a CIR project was sampled and mixed with water to reach the optimum moisture content. The wetted RAP was placed into wood molds having dimensions of 54 in. × 23 in. × 3.25 in. The material was compacted using a plate compactor until a density of approximately 123 lbs/ft3 was achieved. For the testing conducted in the field, a fine-graded 100% RAP mixture was placed using a paver and compacted with a 12-ton double steel drum roller in an effort to assess the potential of using 100% RAP as a surfacing alternative to treat unsurfaced rural roads. The material was rolled for approximately seven to nine vibratory passes, but no density information from the field trial was available. No standard deviation was calculated for the SPRT torque value during Recommended Test Mean Pooled Standard Deviation Threshold Value (Average of Three Tests) Short-pin raveling test Number of blows 8.4 0.8 7.1 Torque, ft-lb 24.3 2.5 20.2 Long-pin shear test Number of blows 22.8 2.1 19.3 Torque, ft-lb 76.4 8.2 62.9 Table 3.80. Recommended test mean, pooled standard deviation, and threshold value. Test Threshold Value CA SR 22 CIR F-C (Single Observation) MN Cell 1-LD FDR E-C (Average of Three Replicates) 2 Hours Curing 6 Hours Curing 24 Hours Curing 3 Hours Curing Short-pin raveling test Number of blows 7.1 3.7 5.7 11.0 5.6 Torque, ft-lb 20.2 15.4 20.2 30.2 19.7 Long-pin shear test Number of blows 19.3 14.3 15.7 26.0 13.1 Torque, ft-lb 62.9 67.0 78.9 123.7 60.0 Table 3.81. Threshold values and Phase III results from two sites.

113 Table 3.82. Results of laboratory and field trials using 0% binder material. Test Threshold Value Laboratory Trial Field Trial Mean Pooled Standard Deviation Zero-Binder Upper Limit (Mean + 1.645 Standard Deviations) Mean Pooled Standard Deviation Zero-Binder Upper Limit (Mean + 1.645 Standard Deviations) Short-pin raveling test Number of blows 7.1 3.7 0.8 5.0 2.4 0.8 3.7 Torque, ft-lb 20.2 12.6 2.5 15.6 15.4 2.5 19.5 Long-pin shear test Number of blows 19.3 7.3 2.1 9.8 6.2 2.1 9.7 Torque, ft-lb 62.9 42.2 8.2 51.9 26.1 8.2 39.6 the field trial since the test value was near the lower limit of the torque wrench. When evaluating the 0% binder trial test results with respect to the threshold values, a direct comparison between the mean values of the test results and the threshold values was not made; rather, a more conservative approach was undertaken to eliminate the probability of accepting a defi- cient material. Given that the 0% binder trials were to repre- sent a worst-case scenario, it was decided that the upper end of the distribution of test results, assumed to be at the 95th percentile, should be less than the threshold value (since the threshold value was considered a minimum acceptable test result). The zero-binder upper limit for both the laboratory and field trials was then calculated by adding 1.645 times the pooled standard deviation to the mean test value, as is shown in Table 3.82. The pooled standard deviation was used since there were more observations (50) in the study population than the zero-binder trials (three). The zero-binder upper limits from the laboratory and field trials were found to be less than the threshold values from each test. The zero- binder trials were considered an acceptable verification of the threshold values and indicated a good level of conserva- tism and a low risk that rejectable material would be found adequate. 3.7 Proposed AASHTO Standard Practice A proposed AASHTO standard practice guide was devel- oped to assist agencies with implementing the proposed tests. A draft of the proposed standard practice is provided in Appendix B.

Next: Chapter 4 - Conclusions and Suggested Research »
Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Pavement recycling is a technology that can restore the service life of pavement structures and stretch available funding for pavement rehabilitation. In general, pavement recycling techniques remix the existing pavement material and reuse it in the final pavement in the form of a stabilized layer.

Limitations to further widespread implementation of pavement recycling processes have been reported in previous national research efforts. The TRB National Cooperative Highway Research Program's NCHRP Research Report 960: Proposed AASHTO Practice and Tests for Process Control and Product Acceptance of Asphalt-Treated Cold Recycled Pavements investigates and recommends a series of tests that could be used for the purpose of implementing rapid quality tests that can be used to assess the time to opening to traffic and time to surfacing a newly constructed recycled layer.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!