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

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

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

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

29   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 Source: Utah DOT 2017. final compaction in addition to a release to traffic. Figure 3.2.  Utah DOT shear vane used for cold VanFrank (2015) also noted that the DCP was used as a recycled materials. 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 tively. This modified shear vane is much more robust than traffic. It is not yet clear why the agency did not include DCP similar products used for unbound materials and is made testing in the instruction manual. However, VanFrank (2015) from a 3-in.-square steel washer having a 3⁄16 in. thickness, stated that during field testing, it was found that both the a 5⁄8-in.-diameter bolt that has the end fashioned to a point, shear vane and DCP criteria could be met while the recycled and 1⁄8-in.-thick steel plate flanges welded to the bolt and the material was still in a plastic state—that is, still susceptible steel washer. The shear vane is hammered into a CIR layer to flow instability. The author attributed the inability of the using a 5-lb hammer until the top washer is flush with the shear vane to identify this as because of the localized nature surface. A torque wrench is attached to the bolt head using a of the testing and the high degree of particle displacement standard socket, and torque is applied such that the end of the around the shear vane edges. The author found that including torque wrench travels 90° in 10 seconds. The greatest torque the MH field test allowed assessment of the bulk particle read on the dial of the torque wrench prior to the material movement from a larger stress influence. VanFrank (2015) breaking loose is recorded as the shear value (in foot-pounds) also noted that multiple tests were needed to achieve the along with the pavement temperature at a depth of 2 in. desired result of identifying a time for release to traffic. VanFrank (2015) stated that the recycled layer was ready for traffic when a shear value of 30 ft-lb was obtained during 3.1.2  Stakeholder Survey field testing. 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. Source: Utah DOT 2017. Figure 3.6 shows the geographic location of the survey respon- Figure 3.3.  Schematic of Utah DOT shear vane used dents; 44% of the respondents reported their location as for cold recycled materials. either northeast or northcentral United States.

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

31   30% Foam 25% Emulsion Percentage of Responses 20% 15% 10% 5% 0% FDR CIR CCPR Recycling Process Figure 3.7.  Stabilizing/recycling agent by recycling process. 70% Foam 60% Emulsion 50% Percentage of Responses 40% 30% 20% 10% 0% Cement Lime Others Chemical Additive/Active Filler Figure 3.8.  Active fillers used with FDR. 70% Foam 60% Emulsion 50% Percentage of Responses 40% 30% 20% 10% 0% Cement Lime Others Chemical Additive/Active Filler Figure 3.9.  Active fillers used with CIR.

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

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

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

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   100% 94% Percent of Specifications withTest 80% 67% 64% 60% 60% Requirement 40% 22% 16% 20% 13% 6% 2% 0% ITS DCP Density Curing Moisture Marshall Raveling Test/Proof Gradation Time Content Stability Stability Rolling Test Type Figure 3.15.  Distribution of constructed quality characteristic including all recycling processes. 100% 92% Percent of Specifications withTest 80% 71% 63% Requirement 60% 46% 46% 40% 29% 20% 4% 0% 0% 0% ITS DCP Density Curing Moisture Marshall Raveling Test/Proof Gradation Time Content Stability Stability Rolling Test Type Figure 3.16.  Distribution of constructed quality characteristic for FDR. 100% 96% Percent of Specifications withTest 80% 69% 67% 64% Requirement 60% 40% 20% 18% 20% 7% 4% 2% 0% ITS DCP Density Curing Moisture Marshall Raveling Test/Proof Gradation Time Content Stability Stability Rolling Test Type Figure 3.17.  Distribution of constructed quality characteristic for CIR.

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

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

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

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

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

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

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

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

46 3.4.3.1  Soil Stiffness Gauge Table 3.12.  Test, curing time, and slab set Figure 3.24 shows the average stiffness of the mixtures information. measured using the SSG at 2 and 72 hours of curing. (For clarity, error bars are not shown.) The average value is Test Curing Time (hours) made up of three tests per replicate, with the number of Slab Set 1 Slab Set 2 Slab Set 3 Moisture 2, 72 replicates shown in Table 3.13. As seen in Figure 3.24, the Soil stiffness gauge 2, 72 SSG was generally able to capture the effect of curing time. Lightweight deflectometer 2, 72 1, 3, 6, 24 Of 16 mixtures, three mixtures (Mixtures 8, 13, and 14) Dynamic cone penetrometer 2, 72 1, 3, 6, 24 Marshall hammer showed a decrease in stiffness with respect to curing time. Long-pin shear test 1, 3, 6, 24 In addition, when individual specimens were considered, Short-pin raveling test 1, 3, 6, 24 10 of 30 specimens had a lower stiffness with respect to Table 3.13.  Specimen and mixture details for moisture content testing. Active Agent Actual Mix Stabilizing/Recycling Active Filler No. of Process State Content, Density, ID Agent Filler Content, Replicates % pcf % 1 IN 2.5 1.0 119.6 2 CCPR 2 VA 2.5 1.0 131.5 3 Cement 3 TX 4.5 1.1 122.9 1 FDR 4 CA 2.5 1.0 127.8 0 124.8 2 5 Emulsified asphalt NY 3.0 0.0 CCPR 130.2 (2 densities) 6 No VA 2.5 0.0 128 2 7 cement CIR ON 1.2 0.0 120.5 2 8 IN 2.5 0.0 118 2 FDR 9 CA 2.5 0.0 127.8 0 10 CCPR VA 2.5 1.0 127.6 0 11 CA 2.0 1.0 120.4 2 CIR 12 Cement MA 2.5 1.0 119.4 2 13 TX 2.4 1.5 125.9 2 FDR 14 CA 2.5 1.0 126.6 3 Foamed asphalt 15 CCPR VA 2.5 0.0 127.6 0 16 MI 2.2 0.0 129.8 2 No CIR 121.3 2 17 cement WI 2.0 0.0 118.6 (2 densities) 18 FDR CA 2.5 0.0 127.8 3 7000 6800 y = 96.42x + 5,988.54 R² = 0.97 Troxler Frequency, Hz 6600 6400 6200 6000 5800 0 1 2 3 4 5 6 7 8 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.

47   2-hour 72-hour 8500 Emulsion, Emulsion, No Foam, No Cement Cement Foam, Cement Cement 8000 Frequency, Hz 7500 7000 6500 6000 0 2 4 6 8 10 12 14 16 18 20 Mixture ID Figure 3.20.  Electromagnetic moisture device results for all mixtures. 10 Emulsion, Emulsion, No Foam, No Cement Foam, Cement Cement Cement 8 Coefficient of Variation, % 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 Mixture ID Figure 3.21.  Within-test slab variability for electromagnetic moisture device based on replicate measurements. 2-hour 72-hour 10 Emulsion, Emulsion, No Foam, No Foam, Cement Cement Cement Cement 8 Coefficient of Variation, % 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 Mixture ID Figure 3.22.  Between-test slab variability for electromagnetic moisture device based on replicate specimens.

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

49   Table 3.14.  Descriptive statistics of SSG stiffness by curing time. Curing Mean, Minimum, Quartile 1, Quartile 3, Maximum, Range, Interquartile Time MN/m MN/m MN/m MN/m MN/m MN/m 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 3.15.  Descriptive statistics of SSG stiffness by recycling agent and active filler type. Interquartile Material Mean, Minimum, Quartile Quartile Maximum, Range, Range, Combination MN/m MN/m 1, MN/m 3, MN/m MN/m MN/m 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 have a larger influence on the SSG-measured stiffness of points in Figure 3.26 are due to not having a replicate speci- certain mixtures. Further, when the mean SSG stiffness men for a given mixture. In general, the COV among mixtures values were compared, those mixtures that included foamed was less than 30% (30 of 32 conditions). The high variability asphalt (both with and without cement as an active filler) seen for Mixture 13 was also observed in other tests consid- tended to be stiffer than those mixtures that included emul- ered in this study, as shown in the following sections. The sified asphalt; the difference was greater for those mixtures average COV was 14.3% and 18.2% for the 2- and 72-hour that included cement. testing, respectively. The higher COV at 72 hours was due, The within-specimen variability of the SSG stiffness was in part, to the high variability observed for Mixture 13. evaluated in terms of a COV. As shown in Figure 3.25, the A way to evaluate the discrimination potential of the SSG COV for replicate measurements was generally less than 10% test with respect to curing time is to assess the curing ratio. (for 54 of 60 conditions; 30 test specimens at two different The curing ratio is defined here as the ratio of the stiffness curing times). The average COV was 4% and 6% for the 2- and at 72 hours to the stiffness at 2 hours. Figure 3.27 shows the 72-hour tests, respectively. Replicate measurements at each curing ratio of the mixtures. The SSG measurements indi- curing time were possible only for the SSG and LWD tests. cated that a total of three mixtures had a lower stiffness with The variability of the SSG stiffness among mixture repli- an increase in curing time (i.e., a curing ratio of less than 1). cates was also assessed via the COV. Figure 3.26 presents the The curing ratio ranged from 0.69 to 5.36, with an average range of COV among the evaluated mixtures. The missing curing ratio of 1.44. 25 2-hour 20 Coefficient of Variation, % 72-hour 15 10 5 0 0 4 8 12 16 20 24 28 32 Specimens Figure 3.25.  Within-specimen soil stiffness gauge variability in terms of a coefficient of variation.

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

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

52 Table 3.17.  Descriptive statistics of LWD modulus by curing time. Curing Mean, Minimum, Quartile 1, Quartile 3, Maximum, Range, Interquartile Time ksi ksi ksi ksi ksi ksi 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 3.18.  Descriptive statistics of LWD modulus by recycling agent and active filler type. Material Mean, Minimum, Quartile 1, Quartile 3, Maximum, Range, Interquartile Combination ksi ksi ksi ksi ksi ksi 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 25 2-hour 20 Coefficient of Variation, % 72-hour 15 10 5 0 0 4 8 12 16 20 24 28 32 Specimens Figure 3.29.  Within-specimen LWD variability in terms of a coefficient of variation. 2-hour 72-hour 50 Emulsion, Emulsion, Foam, Foam, Cement No Cement Cement No Cement 40 Coefficient of Variation, % 30 20 10 0 0 2 4 6 8 10 12 14 16 18 20 Mixture ID Figure 3.30.  LWD variability among mixture replicates in terms of a coefficient of variation.

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

54 Table 3.20.  Comparison of SSG and LWD tests. Range/Observation Desired Parameter SSG LWD Trend 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 0.9 to 7.8 (average = 1.44) (average = 1.8) Higher Number of mixtures with statistically significant 1 3 Higher difference between the 2-hour vs. 72-hour tests 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 3.21.  Test slab details for LWD stiffness testing at 1, 3, 6, and 24 hours. Active Agent Actual Mix Stabilizing/Recycling Active Filler No. of Process State Content, Density, ID Agent Filler Content, Replicates % pcf % 3 rep full 1 IN 2.5 1.0 119.1 0 rep half CCPR 2 reps full 2 VA 2.5 1.0 127.6 0 rep half Cement 2 rep full 3 TX 0.5 1.1 131.5 1 rep half FDR 3 reps full 4 CA 2.5 1.0 127.8 1 rep half 3 rep full 5 Emulsified asphalt NY 3.0 0.0 122.0 0 rep half CCPR 5 reps full 6 VA 2.5 0.0 127.6 0 rep half No 2 rep full 7 CIR ON 1.2 0.0 121.4 cement 1 rep half 3 rep full 8 IN 2.5 0.0 119.1 0 rep half FDR 4 reps full 9 CA 2.5 0.0 127.8 0 rep half 5 reps full 10 CCPR VA 2.5 1.0 127.6 1 rep half 3 rep full 11 CA 2.0 1.0 117.4 1 rep half CIR 3 rep full 12 Cement MA 2.5 1.0 121.0 1 rep half 4 reps full 13 TX 2.4 1.5 125.6 1 rep half FDR 3 reps full 14 Foamed asphalt CA 2.5 1.0 127.8 1 rep half 2 reps full 15 CCPR VA 2.5 0.0 127.6 1 rep half 3 rep full 16 MI 2.2 0.0 129.8 No 0 rep half CIR cement 3 rep full 17 WI 2.0 0.0 121.3 0 rep half 2 reps full 18 FDR CA 2.5 0.0 127.8 1 rep half

55   1-hour 3-hour 6-hour 24-hour 50 Emulsion, Emulsion, No Foam, Foam, No Cement Cement Cement Cement 40 LWD Modulus, ksi 30 20 10 0 0 2 4 6 8 10 12 14 16 18 20 Mixture ID Figure 3.32.  Modulus of mixtures as measured by LWD at shorter time intervals. in curing time when the LWD moduli of all mixtures were cement, in general, a continued increase in the LWD modulus combined without consideration of the specific characteristics with respect to an increase in curing time was observed, of the mixtures. Likewise, the spread of the LWD modulus as expected. (as quantified by range and IQR), in general, also increased Figure 3.33 presents the mixture-to-mixture variability with an increase in curing time. Figure 3.32 also reveals that of the LWD modulus measured at shorter curing intervals the impact of curing time on the LWD modulus is more in terms of the COV. The COV for the LWD measurements, evident with the presence of active filler (cement). Table 3.23 considering all curing times, was less than 30% except for a reflects that the spread of the LWD modulus is wider for the few observations (4 of 72). It is interesting that for the curing mixtures with cement than for the mixtures without cement. time considered in this part of the study, two of four obser- It was also noted from the table that the mixtures with cement vations with more than 30% COV were for Mixture 13. With tended to have a higher LWD modulus, as expected. regard to the LWD data for 2- and 72-hour tests in addition Figure 3.32 reveals that there was, generally, a reduction to observations from other tests used in this study, Mixture 13 in the measured LWD modulus followed by a slight increase had a consistently higher variability compared to the other in modulus over the four curing periods for the mixtures mixtures. Nevertheless, the overall average COV for the LWD without cement, an observation also noted with other tests measurements at shorter curing intervals was 15.9%, which considered in this study. For those mixtures incorporating was slightly higher than the average COV of 14.9% and 15.2% Table 3.22.  Descriptive statistics of LWD modulus by curing time. Curing Mean, Minimum, Quartile Quartile Maximum, Range, Interquartile Time ksi ksi 1, ksi 3, ksi ksi ksi 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 3.23.  Descriptive statistics of LWD modulus by recycling agent and active filler type. Material Mean, Minimum, Quartile 1, Quartile Maximum, Range, Interquartile Combination ksi ksi ksi 3, ksi ksi ksi 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

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

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

58 8.0 10 Blows, 2 hrs Curing 7.0 10 Blows, 72 hrs Curing Average Penetrated Depth, mm 6.0 5.0 4.0 3.0 2.0 1.0 0.0 1 2 3 5-1 5-2 6 7 8 11 12 13 14 16 17-1 17-2 18 Mixture ID Figure 3.36.  Average penetrated depths at 10 MH blows after 2- and 72-hour curing. 8.0 15 Blows, 2 hrs Curing 7.0 15 Blows, 72 hrs Curing Average Penetrated Depth, mm 6.0 5.0 4.0 3.0 2.0 1.0 0.0 1 2 3 5-1 5-2 6 7 8 11 12 13 14 16 17-1 17-2 18 Mixture ID Figure 3.37.  Average penetrated depths at 15 MH blows after 2- and 72-hour curing. 8.0 20 Blows, 2 hrs Curing 7.0 20 Blows, 72 hrs Curing Average Penetrated Depth, mm 6.0 5.0 4.0 3.0 2.0 1.0 0.0 1 2 3 5-1 5-2 6 7 8 11 12 13 14 16 17-1 17-2 18 Mixture ID Figure 3.38.  Average penetrated depths at 20 MH blows after 2- and 72-hour curing.

59   Table 3.25.  Descriptive statistics of MH testing by number of blows and curing time. Penetrated Depth, mm Number of Curing Interquartile Blows Time Mean Minimum Quartile 1 Quartile 3 Maximum Range Range 2 hours 1.22 0.46 0.66 1.44 3.33 2.87 0.78 5 72 hours 0.58 0.26 0.39 0.76 1.19 0.93 0.37 2 hours 1.90 0.76 1.13 2.21 5.64 4.88 1.08 10 72 hours 1.00 0.44 0.67 1.47 1.67 1.23 0.79 2 hours 2.49 1.03 1.54 3.00 6.81 5.78 1.46 15 72 hours 1.28 0.53 0.96 1.72 2.27 1.74 0.76 2 hours 3.06 1.57 1.98 3.92 7.89 6.32 1.94 20 72 hours 1.51 0.66 1.03 1.92 2.67 2.01 0.89 Table 3.25 shows the descriptive statistics of the pene­ COV values were lower for nearly all mixtures at 20 blows trated depth with respect to number of MH blows (i.e., when compared with a lesser number of blows at both 2 and five, 10, 15, and 20) and curing time (i.e., 2 and 72 hours). 72 hours. The average penetrated depth COV at five, 10, 15, The mean penetrated depth increased with respect to the and 20 blows was less at 2 hours after fabrication than at number of blows and decreased with respect to curing 72 hours after fabrication. At 2 hours after fabrication, seven, time, as expected. Similarly, the IQR increased with respect eight, and seven mixtures had COVs of less than 30% at 10, to the number of blows and decreased with respect to 15, and 20 blows, respectively. curing time. The generated data were further analyzed to investigate The variability of the MH test results at the recorded blow the effect of multiple parameters on the measured MH counts was evaluated in terms of a COV computed for those penetrated depth. An ANCOVA at a confidence level of 95% mixtures that had replicates (11 of 16 mixtures). Figures 3.39 was used to evaluate the significance of these parameters. and 3.40 show the penetrated depth COV (averaged across Table 3.26 presents the outcomes of the ANCOVA. The agent multiple slabs) from all evaluated mixtures at 2 and 72 hours rate factor was nested within the recycling agent factor. after fabrication, respectively, at five, 10, 15, and 20 MH Moreover, the cement content factor was nested within the blows. The COV decreased with an increasing number of recycling agent factor. The analysis shows that all factors, blows. The research team attributed this to a reduced influ- except the process type, significantly affected the MH pene- ence of surface texture as the penetrated depth increased. trated depth (i.e., p-value < 0.05). 5 blows - 2 hrs Curing 10 blows - 2 hrs Curing 15 blows - 2 hrs Curing 20 blows - 2 hrs Curing 100 Emulsion, Emulsion, Foam, Foam, 90 Cement No Cement Cement No Cement Coefficient of Variation, % 80 70 60 50 40 30 20 10 0 1 2 6 7 8 11 12 13 14 16 18 Mixture ID Figure 3.39.  Penetrated depth variability after 2-hour curing duration in terms of coefficient of variation.

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

61   2-hour 72-hour 12 Emulsion, Emulsion, Foam, Foam, Cement No Cement Cement No Cement 10 8 DPI, mm/blow 6 4 2 0 1 2 3 4 5-1 5-2 6 7 8 9 10 11 12 13 14 15 16 17-1 17-2 18 Mixture ID Figure 3.41.  DPI for all evaluated mixtures. Table 3.27.  Descriptive statistics of DPI with respect to curing time. DPI, mm/blow Curing Time Mean 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 3.28.  Descriptive statistics for DPI with respect to recycling agent and active filler type. DPI, mm/blow Material Interquartile Combination Mean Minimum Quartile 1 Quartile 3 Maximum Range Range Emulsion, cement 3.1 1.7 1.9 4.6 4.8 3.1 2.7 Emulsion, no 4.9 2.5 3.7 7.0 7.9 5.4 3.3 cement 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 2-hour 72-hour 50 Emulsion, Emulsion, Foam, Foam, Cement No Cement Cement No Cement 40 Coefficient of Variation, % 30 20 10 0 1 2 3 4 5-1 5-2 6 7 8 9 10 11 12 13 14 15 16 17-1 17-2 18 Mixture ID Figure 3.42.  Coefficient of variation for DPI for mixtures with replicates.

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

63   Table 3.30.  Test slab details for long-pin shear test at 1, 3, 6, and 24 hours. Active Agent Actual Mix Stabilizing/Recycling Active Filler No. of Process State Content, Density, ID Agent Filler Content, Replicates % pcf % 1 IN 2.5 1.0 119.1 2 CCPR 2 VA 2.5 1.0 127.6 0 Cement 3 TX 4.5 1.1 131.5 2 FDR 4 CA 2.5 1.0 127.8 1 5 Emulsified asphalt NY 3.0 0.0 122.0 2 CCPR 6 VA 2.5 0.0 127.6 3 No 7 CIR ON 1.2 0.0 121.4 2 cement 8 IN 2.5 0.0 119.1 2 FDR 9 CA 2.5 0.0 127.8 2 10 CCPR VA 2.5 1.0 127.6 3 11 CA 2.0 1.0 117.4 2 CIR 12 Cement MA 2.5 1.0 121.0 2 13 TX 2.4 1.5 125.6 2 FDR 14 Foamed asphalt CA 2.5 1.0 127.8 3 15 CCPR VA 2.5 0.0 127.6 0 16 No MI 2.2 0.0 129.8 2 CIR 17 cement WI 2.0 0.0 121.3 2 18 FDR CA 2.5 0.0 127.8 0 1-hour 3-hour 6-hour 24-hour 100 Emulsion, Emulsion, No Foam, No Cement Foam, Cement 80 Cement Cement Number of Blows 60 40 20 0 0 2 4 6 8 10 12 14 16 18 Mixture ID Figure 3.44.  Number of blows to drive shear fixture into laboratory produced slabs. 1-hour 3-hour 6-hour 24-hour 300 Emulsion, Emulsion, Foam, Foam, No 250 Cement No Cement Cement Cement 200 Torque, ft-lbs 150 100 50 0 0 2 4 6 8 10 12 14 16 18 Mixture ID Figure 3.45.  Torque values for field long-pin shear test.

64 Table 3.31.  Descriptive statistics of number of blows by curing time. 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 3.32.  Descriptive statistics of number of blows by recycling agent and active filler type. Material Interquartile Mean Minimum Quartile 1 Quartile 3 Maximum Range Combination 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 The descriptive statistics of the LPST testing are shown and foamed asphalt with cement. The IQR was similar for in Tables 3.31 through 3.34. Tables 3.31 and 3.32 show the emulsified mixtures with and without cement but decreased descriptive statistics of the number of blows by curing time for foamed asphalt mixtures without cement. and recycling agent type, respectively. Tables 3.33 and 3.34 The variability of the LPST was assessed in terms of the show the descriptive statistics of the torque value by curing COV calculated from testing replicate specimens. Figure 3.46 time and recycling agent type, respectively. Table 3.31 shows shows the COV for the number of blows, and Figure 3.47 that the mean number of blows and the IQR increased with shows the COV for the measured torque values. The COV respect to curing time. Table 3.32 shows that the mean number for the number of blows was generally less than 20% (46 of of blows was greater when cement as an active filler was 54 conditions, considering all curing times), with an average present for mixtures having both emulsified and foamed COV of 9.5%. Similarly, the measured torque value COV was asphalt. The IQR was less when cement was included as generally less than 20% (47 of 49 conditions), with an average an active filler for mixtures using emulsified asphalt but COV of 9%. Although some mixtures had a COV of greater greater for mixtures using foamed asphalt. Table 3.33 shows than 20%, the data suggested that the variability was not par- that the mean torque value increased with respect to curing ticularly affected by the process type, recycling agent type, time, as expected. As noted previously, for some mixtures, or curing time. In general, those mixtures that had lower or the torque value at 24 hours could not be recorded since the higher COVs at the 1-hour test tended to have a relatively value exceeded the maximum capacity of the handheld similar variability across all curing times. torque wrench. The IQR decreased from 1 to 3 hours but then The generated data were also analyzed statistically to increased from 3 to 24 hours. Table 3.34 shows that the investigate the effect of the recycled mixture parameters torque values increased for mixtures using both emulsified considered in this study on the measured torque value and Table 3.33.  Descriptive statistics of torque values by curing time. Mean, Minimum, Quartile Quartile Maximum, Range, Interquartile Curing Time ft-lbs ft-lbs 1, ft-lbs 3, ft-lbs ft-lbs ft-lbs 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 3.34.  Descriptive statistics of torque values by recycling agent and active filler type. Material Mean, Minimum, Quartile Quartile Maximum, Range, Interquartile Combination ft-lbs ft-lbs 1, ft-lbs 3, ft-lbs ft-lbs ft-lbs 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 79.5 30.8

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

66 Table 3.35.  Results of ANCOVA Table 3.36.  Results of ANCOVA for number of blows. for torque value. Source DF f-Value p-Value Source DF f-Value p-Value Slab density 1 32.89 0.000 Slab density 1 36.75 0.000 Recycling process 1 22.17 0.000 Recycling process 1 10.56 0.002 Recycling agent type 1 2.3 0.132 Recycling agent type 1 17.64 0.000 Curing time 3 20.38 0.000 Curing time 3 6.44 0.001 Recycling agent content 6 10.38 0.000 Recycling agent content 6 5.57 0.000 Note: DF = degrees of freedom; bold/highlight = p-value Note: DF = degrees of freedom; bold/highlight = p-value shows the source to be significant. shows the source to be significant. were produced, whereas one replicate was produced for other pin to the tips of the shorter outer pins. The value N2 mixtures. Table 3.37 also shows that some replicates were describes the number of blows required to drive the raveling produced using the full design stabilizing/recycling agent fixture from the tip of the long center pin until the base content, and some were produced at half the design content plate was seated against the surface of the test slab. These in an effort to force a more severe raveling situation. two separate values were recorded in case they proved useful The raveling data are presented with respect to two mea- in the analysis. For each of these measurements, the data surements of the number of blows (N1 and N2) required to are presented at the full design binder content and for certain drive the raveling fixture into the test slab and the measured mixtures at half of the design binder content. torque. The value N1 describes the number of blows required Figures 3.48 and 3.49 show the number of blows (N1) at to drive the raveling fixture from the tip of the longer center the full and half design binder contents, respectively. The Table 3.37.  Specimen and mixture details for short-pin raveling test. Mix Agent Filler Actual No. of Agent Filler Process State ID Content, % Content, % Density, pcf Replicates 1 rep full 1 IN 2.5 1.0 119.1 0 rep half CCPR 2 reps full 2 VA 2.5 1.0 127.6 0 rep half Cement 1 rep full 3 TX 4.5 1.1 131.5 1 rep half FDR 2 reps full 4 CA 2.5 1.0 127.8 1 rep half Emulsified 1 rep full 5 NY 3.0 0.0 122.0 asphalt 0 rep half CCPR 2 reps full 6 VA 2.5 0.0 127.6 0 rep half No 1 rep full 7 CIR ON 1.2 0.0 121.4 cement 1 rep half 1 rep full 8 IN 2.5 0.0 119.1 0 rep half FDR 2 reps full 9 CA 2.5 0.0 127.8 0 rep half 2 reps full 10 CCPR VA 2.5 1.0 127.6 1 rep half 1 rep full 11 CA 2.0 1.0 117.4 1 rep half CIR 1 rep full 12 Cement MA 2.5 1.0 121.0 1 rep half 2 reps full 13 TX 2.4 1.5 125.6 1 rep half FDR Foamed 2 reps full 14 CA 2.5 1.0 127.8 asphalt 1 rep half 2 reps full 15 CCPR VA 2.5 0.0 127.6 1 rep half 1 rep full 16 MI 2.2 0.0 129.8 No 0 rep half CIR cement 1 rep full 17 WI 2.0 0.0 121.3 0 rep half 2 reps full 18 FDR CA 2.5 0.0 127.8 1 rep half

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

68 Table 3.38.  Descriptive statistics of SPRT parameters for full and half binder specimens irrespective of curing time, recycling agent type, and cement content. Interquartile Parameter Mean Minimum Quartile 1 Quartile 3 Maximum Range 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 3.39.  Descriptive statistics of number of blows (N1) for full and half binder specimens with respect to curing time. N1, Blows Curing Time Mean 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 3.40.  Descriptive statistics of number of blows (N1) for full and half binder specimens with respect to recycling agent type and cement content. N1, Blows Material Interquartile Combination Mean Minimum Quartile 1 Quartile 3 Maximum Range 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 1-hour 3-hour 6-hour 24-hour 120% Emulsion, Emulsion, Foam, Foam, COV for Number of Blows (N1), % 100% Cement No Cement Cement No Cement 80% 60% 40% 20% 0% 0 2 4 6 8 10 12 14 16 18 Mixture ID Figure 3.50.  Coefficient of variation for number of blows, N1 (full binder specimens only).

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

70 Table 3.41.  Descriptive statistics of number of blows (N2) for full and half binder specimens with respect to curing time. N2, Blows Curing Time Mean 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 3.42.  Descriptive statistics of number of blows (N2) for full and half binder specimens with respect to recycling agent type and cement content. N2, Blows Material Interquartile Combination Mean Minimum Quartile 1 Quartile 3 Maximum Age 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 1-hour 3-hour 6-hour 24-hour 100% Emulsion, Emulsion, Foam, Foam, COV for Number of Blows (N2), % Cement No Cement Cement No Cement 80% 60% 40% 20% 0% 0 2 4 6 8 10 12 14 16 18 Mixture ID Figure 3.53.  Coefficient of variation for number of blows, N2 (full binder specimens only). 1-hour 3-hour 6-hour 24-hour 120 Emulsion, Emulsion, No Foam, No Foam, Cement 100 Cement Cement Cement 80 Torque, ft-lbs 60 40 20 0 0 2 4 6 8 10 12 14 16 18 Mixture ID Figure 3.54.  Torque value using raveling fixture (full binder content).

71   1-hour 3-hour 6-hour 24-hour 120 Emulsion, Emulsion, No Foam, No Foam, Cement 100 Cement Cement Cement 80 Torque, ft-lbs 60 40 20 0 0 2 4 6 8 10 12 14 16 18 Mixture ID Figure 3.55.  Torque value using raveling fixture (half binder content). blows, Figure 3.55 shows that the torque values were similar were all less than about 25%. There did not appear to be a at the half binder and at the full binder content. clear trend based on material type, partly because of the low The descriptive statistics for the measured torque value number of mixtures that had replicates. using all collected data with respect to curing time are shown An ANCOVA at a confidence level of 95% was used to in Table 3.43. The mean torque value increased with respect evaluate the significance of various mixture parameters. The to curing time. The IQR also increased with curing time agent rate and cement type factors were both nested within from 1 to 6 hours but decreased from 6 to 24 hours. Table 3.44 the recycling agent type factor. Tables 3.45 through 3.47 shows the descriptive statistics for the measured torque value present the outcomes of the ANCOVA for N1, N2, and ravel- using all collected data with respect to material combina- ing torque, respectively. The p-values in Table 3.45 show tions. The mean torque value increased with the presence that N1 was sensitive only to process type factor. Table 3.46 of cement for mixtures using both emulsified and foamed shows that the number of blows (N2) was sensitive to all asphalt. A similar increasing trend was observed for the IQR. evaluated factors. Table 3.47 shows that the measured torque Figure 3.56 shows the variability of the torque value in value was sensitive to all factors except the recycling process terms of the COV. The COV values were considered low and and the recycling agent type. Table 3.43.  Descriptive statistics of torque value using raveling fixture for full and half binder specimens with respect to curing time. Raveling Torque, ft-lb Curing Time Mean 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 3.44.  Descriptive statistics of torque value using raveling fixture for full and half binder specimens with respect to recycling agent and active filler type. Raveling Torque, ft-lb Material Interquartile Combination Mean Minimum Quartile 1 Quartile 3 Maximum Range 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

72 1-hour 3-hour 6-hour 24-hour 100% Emulsion, Emulsion, Foam, Foam, No Cement No Cement Cement Cement 80% COV for Torque, % 60% 40% 20% 0% 0 2 4 6 8 10 12 14 16 18 Mixture ID Figure 3.56.  Coefficient of variation for torque value (full binder specimens only). Table 3.45.  Short-pin raveling test: 3.4.8  Correlation Analysis results of ANCOVA for number of blows, N1. A correlation analysis was performed to investigate the relationship between selected test measurement combina- Parameter DF f-Value p-Value tions using the data from the Phase II laboratory study. The Slab density 1 24.09 0.000 Recycling process 1 6.37 0.014 analysis was performed by calculating the Pearson correlation Recycling agent type 1 4.14 0.046 coefficient (r) and the associated p-value. The Pearson cor- Curing time 3 9.61 0.000 relation coefficient describes the linear relationship between Note: DF = degrees of freedom; bolding indicates two variables and has a range of −1 < r < +1, where values that the p-value shows the source to be significant. closer to −1 or +1 indicate a stronger correlation. A value of −1 or +1 indicates a negative or positive relationship, respec- Table 3.46.  Short-pin raveling test: tively. The p-value indicates the statistical significance of the results of ANCOVA for number relationship; a higher p-value suggests that the correlation of blows, N2. may be due to random chance. Tables 3.48 and 3.49 show the Pearson correlation coef- Parameter DF f-Value p-Value ficient and p-value for comparisons of the test slab density, Slab density 1 33.81 0.000 Recycling process 1 20.29 0.000 SSG stiffness, LWD modulus, and DPI values at curing periods Recycling agent type 1 10.37 0.002 of 2 and 72 hours, respectively. For those combinations that Curing time 3 12.68 0.000 were shown to have a strong correlation (|r| > 0.7, based on Note: DF = degrees of freedom; bolding indicates categories by Evans [1996]), the p-value was determined to that the p-value shows the source to be significant. estimate the significance of the relationship (alpha = 0.05). Shaded cells indicate comparisons where both conditions Table 3.47.  Short-pin raveling test: were met. There was a strong correlation between the SSG results of ANCOVA for torque. and LWD at 2 hours of curing. For both comparisons, the p-value indicates that the correlation was statistically Parameter DF f-Value p-Value significant. Slab density 1 34.08 0.000 Recycling process 1 0.69 0.411 Tables 3.50 through 3.53 show the results of the correlation Recycling agent type 1 4.26 0.000 analysis for those slabs tested at 1, 3, 6, and 24 hours of Curing time 3 5.43 0.002 curing time, respectively. The tables show the Pearson cor- Note: DF = degrees of freedom bolding indicates relation coefficient and p-value for comparisons of the test that the p-value shows the source to be significant. 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   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 −0.1815 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 not included because of the high test variability. For those and N2, respectively, for curing times of 1, 3, 6, and 24 hours. combinations that were shown to have a strong correlation The trendline slopes were similar for all curing times, and (|r| > 0.7), the p-value was determined to estimate the signifi- the coefficient of determination generally increased with cance of the relationship (alpha = 0.05). Shaded cells indicate respect to curing time. Figure 3.59 shows the relationship comparisons where both conditions were met. between the SPRT number of blows N1 and N2 for all curing The number of blows from the LPST and SPRT have a times. The trendline slope values were similar across all strong correlation, and the relationship was statistically signi­ curing times, and the coefficient of determination increased ficant across all four curing times. The LPST torque value with respect to curing time. Figure 3.60 shows the relation- and the SPRT number of blows also had a strong correlation ship between the LPST torque value and the SPRT number with the DPI, with relationships statistically significant at 1 of blows (N2) for all curing times. The trendline slopes were and 24 hours of curing. The SPRT torque value had a strong similar for all curing times, and the coefficient of determina- correlation with the DPI but with the relationship statistically tion generally increased with respect to curing time. significant only at the 24-hour curing time. Slab density did Figure 3.61 shows the relationship between the LPST torque not have a strong correlation with any of the performance value and DPI for curing times of 1 and 24 hours. The coeffi- tests. The LPST torque value did not have a strong correlation cient of determination increased with respect to curing time, with the SPRT torque value. and the slope of the trendline became more negative as curing Figures 3.57 through 3.60 demonstrate the relationship time increased. Figure 3.62 shows the relationship between between those tests shown in the correlation analysis to the SPRT torque value and DPI for both curing times. As with have the strongest correlation and a statistically significant Figure 3.61, the coefficient of determination increased with relationship. The data are presented with respect to the respect to curing time, but the slope of the trendline became curing time, and linear trendlines are shown for each. Another less negative as curing time increased. Figure 3.63 shows the trendline type (e.g., polynomial, exponential) might show a relationship between the SPRT N2 and DPI for both curing higher coefficient of determination, but use of a linear trend times. The coefficient of determination increased slightly is consistent with the relationship shown by the Pearson with respect to curing time, and the slope of the trendline correlation coefficient. became less negative as curing time increased. A nonlinear Figures 3.57 and 3.58 show the relationship between the trendline would likely better describe the relationships shown LPST number of blows and the SPRT number of blows N1 in these three figures. Table 3.49.  Correlation analysis at 72 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

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

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

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

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

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

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

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

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

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

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

84 1.8 Reference Line with 1.6 Slope 1/Seffect Length 1.4 1.2 Tip Dullness Half Normal 1 0.8 Tip Angle 0.6 Torque Angular Rate 0.4 0.2 Outer Pin Diameter 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Effect, Number of Blows (a) 2.5 Reference Line with Slope 1/Seffect 2 Tip Dullness Half Normal 1.5 Torque Angular Rate 1 Pin Length Tip Angle 0.5 Outer Pin Diameter 0 0.0 0.5 1.0 1.5 2.0 2.5 Effect, Torque (b) Figure 3.67.  Half-normal plot for SPRT, (a) number of blows, (b) torque value. Table 3.57.  Statistical significance of factors for LPST and SPRT. Factor Property Outer Pin Test Pin Length, Torque Angular Tip Measure Tip Angle, ° Diameter, in. Rate, °/sec Dullness in. Long-pin shear No. of blows NS NS NA NS S test Torque, ft-lb NS NS S NS S Short-pin No. of blows S NS NA NS NS raveling test Torque, ft-lb NS NS NS S NS NS = not significant; S = significant; NA = not applicable. Table 3.58.  Properties assessed and tests conducted during field testing. 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)

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

86 45 40 35 LWD Modulus, ksi 30 25 20 15 10 5 2 2 3 1 5 3 0 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 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. Figure 3.71 shows that the LWD modulus of the CIR range of material properties. Figure 3.71 shows that the projects was like the FDR and CCPR modulus values. How- support conditions likely influenced the test results. As can ever, two CIR projects showed much higher values, and one be observed in the results for the other tests, those tests that showed much lower values, than the other projects. Two of act only on the recycled layer do not show the same differ- these relatively extreme values occurred from two test sections ence in properties as identified by the LWD modulus for the at the project from Indiana. This project was constructed GS and PS sections at the Indiana project. on a section of roadway that had good foundation material (shown in Figure 3.71 as GS for good support) in the travel 3.5.2  Penetration Resistance lanes but poor quality material (shown in Figure 3.71 as PS for poor support) in the shoulder areas. The research team Figure 3.72 shows the results of DCP field testing for FDR intentionally tested in these two locations to give a wider and CCPR mixtures, and Figure 3.73 shows them for CIR 45 40 35 LWD Modulus, ksi 30 25 20 15 10 5 1 1 3 1 2 1 3 1 2 2 0 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 CA SR 22 NY 30 NY 28 MN Cell IN SR 1- IN SR 1- CA SR 3 4 GS PS 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 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   12 10 8 DPI, mm/blow 6 4 2 2 2 3 2 5 3 0 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 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. mixtures. The DCP penetration index values are similar for Figure 3.73 shows that the DPI also reflects the influence the FDR and CCPR projects except for the FDR section from of two other material properties. The NY 23A field project New Mexico. Despite the low stiffness and modulus values in New York showed that the DPI is sensitive to changes in indicated earlier, the Minnesota FDR sections do not show a curing time. (DPI decreased with increasing curing time, significant decrease in the penetration index. The Minnesota as expected.) In addition, the different support conditions FDR section with a lower density (noted as LD in Figure 3.72) from Indiana showed that the poor support section had a had a higher penetration index than the Minnesota FDR higher DPI value than the good support section, as might be section with a higher density (noted as HD in Figure 3.72), expected if the underlying condition had an influence on as expected. This shows that for similar material, the DCP is the recycled material. (As shown in Table 2.3, the densities sensitive to changes in density under field testing conditions. were the same.) 12 10 8 DPI, mm/blow 6 4 2 1 1 3 1 2 1 3 1 2 2 0 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 CA SR 22 NY 30 NY 28 MN Cell IN SR 1- IN SR 1- CA SR 3 4 GS PS 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 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 number of blows. As with the DCP test results, the SR 22 for CIR projects were similar to or slightly higher than those project from California had the fewest blows as compared to for the FDR and CCPR projects. This is especially true for the the rest of the CIR projects. 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 3.5.3.2  Torque Value from New York showed that the DCP penetration index was Figure 3.76 shows the LPST torque value results for FDR sensitive to changes in curing time, where the penetration and CCPR mixtures, and Figure 3.77 shows them for CIR index decreased with respect to curing time, as expected. mixtures. The figures show a similar range of torque values In addition, the different support conditions from Indiana for all three recycling processes. The lower-density FDR showed that the poor support section had a higher penetra- section from the Minnesota project had a lower torque value tion index value than the good support section, as might be than the higher-density section, as expected. The two tests expected if the underlying condition had an influence on from the NY 23A project showed that the LPST torque value the recycled material. (As shown in Table 2.3, the densities was sensitive to changes in curing. A similar LPST torque were the same.) value was observed for the good and poor support conditions from the Indiana project. 3.5.3  Shear Resistance 3.5.3.1a  Number of Blows 3.5.4  Raveling Resistance Figure 3.74 shows the number of blows from the LPST 3.5.4.1a  Number of Blows results for FDR and CCPR mixtures, and Figure 3.75 shows them for CIR mixtures. The figures show a similar range of Figure 3.78 shows the SPRT number of blows for FDR results when FDR and CCPR are compared with CIR mix- and CCPR mixtures, and Figure 3.79 shows them for CIR tures. In addition, the lower-density section from Minnesota mixtures. The number of blows shown from the field testing had fewer blows than the corresponding higher-density is the same as the number of blows (N2) shown from the section, as expected. Figure 3.75 also shows a relatively wider laboratory testing. From Figure 3.78, the SPRT number of range of test results for the CIR mixtures. Results from the blows showed a ranking of projects similar to that of the NY 23A project indicated that the LPST number of blows LPST number of blows. As expected, the lower-density FDR was sensitive to changes in curing time in the field. The two section from the Minnesota project had fewer blows than the support conditions from the Indiana project showed a similar higher-density FDR section. 50 45 40 LPST Number of Blows 35 30 25 20 15 10 5 2 2 3 3 5 3 0 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 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   50 45 40 LPST Number of Blows 35 30 25 20 15 10 5 1 1 3 1 2 1 3 1 2 2 0 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 CA SR 22 NY 30 NY 28 MN Cell IN SR 1- IN SR 1- CA SR 3 4 GS PS 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 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. 180 160 140 LPST Torque, ft-lb 120 100 80 60 40 20 2 2 3 3 5 3 0 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 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.

90 180 160 140 LPST Torque, ft-lb 120 100 80 60 40 20 1 1 3 1 2 1 3 1 2 2 0 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 CA SR 22 NY 30 NY 28 MN Cell IN SR 1- IN SR 1- CA SR 3 4 GS PS 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 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. 20 18 16 SPRT Number of Blows 14 12 10 8 6 4 2 2 2 3 3 5 3 0 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 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   20 18 16 SPRT Number of Blows 14 12 10 8 6 4 2 1 1 3 1 2 1 3 1 2 2 0 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 CA SR 22 NY 30 NY 28 MN Cell IN SR 1- IN SR 1- CA SR 3 4 GS PS 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 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 3.5.4.2  Torque Value and fewest number of blows from the SPRT fixture were like those identified using the LPST fixture. The number of blows Figure 3.80 shows the SPRT torque values for FDR and from the SPRT increased with respect to curing time, as seen CCPR mixtures, and Figure 3.81 shows them for CIR mix- from the NY 23A project. In addition, there is little difference tures. From Figure 3.80, the SRPT torque values show a in the number of blows from the Indiana CIR project for the relatively wider range of responses and a similar ranking of good and poor support conditions. projects compared to the LPST. Figure 3.80 also shows some 50 45 40 35 SPRT Torque, ft-lb 30 25 20 15 10 5 2 2 3 3 5 3 0 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 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 50 45 40 35 SPRT Torque, ft-lb 30 25 20 15 10 5 1 1 3 1 2 1 3 1 2 2 0 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 CA SR 22 NY 30 NY 28 MN Cell IN SR 1- IN SR 1- CA SR 3 4 GS PS 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 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. differences between the high- and low-density FDR sections • LPST torque value with SPRT number of blows, SPRT from Minnesota, but the differences are not likely to be statis- torque value, and DPI; and tically significant. Figure 3.81 shows that the SPRT torque • SPRT number of blows with SPRT torque value and DPI. values are also like the LPST torque values in terms of the rankings of the three projects with highest torque values. Figures 3.82 through 3.90 demonstrate the relationship The LPST torque values are similar for seven of the 10 projects between those tests shown in the correlation analysis to have shown, but the SPRT torque values have a wider range over the strongest correlation and a statistically significant relation- these same seven projects. The SPRT torque values also show ship based on the field testing. The data are presented along a difference with respect to curing time for the NY 23A project with a linear trendline to be consistent with the linear relation- and a slight difference in the torque values for the high- and ship shown by the Pearson correlation coefficient. For most low-density FDR sections from Minnesota. 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 3.5.5  Correlation Analysis relationships better. A correlation analysis was performed to investigate the A small cluster of data artificially increased the correlation relationship between the tests performed in the Phase III field for certain comparisons (especially related to the LPST and study. The analysis was performed by calculating the Pearson SPRT results for NM US 491, NY SR 23A, and MN Cell 3). correlation coefficient (r) and the associated p-value. If the data for these three projects are removed from the Table 3.59 shows the Pearson correlation coefficient and analysis, only the SPRT torque values and number of blows p-value for comparisons of the field-measured density, SSG were well correlated (i.e., |r|>0.7). Including all data shows stiffness, LWD modulus, LPST number of blows, LPST torque that the LPST blows and torque, the SPRT blows and torque, value, SPRT number of blows (N1 and N2), SPRT torque and DPI were all well correlated. When the data from the value, and DPI values. For those combinations that were three projects were removed, the correlation between the tests shown to have a strong correlation (|r|>0.7,), the p-value was was reduced. determined to estimate the significance of the relationship (alpha = 0.05). Shaded cells indicate comparisons where both 3.5.6  Lessons Learned During Field Testing conditions were met. The analysis showed that the following combinations had a strong, statistically significant correlation: Prior to conducting any of the tests in the field, suitable and uniform sites were selected based on visual observation • SSG stiffness with LWD modulus; of the recycling process and the completed recycled layer. • LPST number of blows with LPST torque value, SPRT As an example, cement as an active filler was observed to torque value, and DPI; be applied non-uniformly across the width of the lane on

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

94 50 40 LWD Modulus, ksi 30 20 10 y = 1.2738x - 8.6304 R² = 0.8293 0 0 10 20 30 40 50 SSG Stiffness, MN/m Figure 3.82.  Relationship between soil stiffness gauge stiffness and LWD modulus. 180 160 140 120 LPST Torque, ft-lb 100 80 60 40 y = 2.4868x + 20.0703 20 R² = 0.7813 0 0 10 20 30 40 50 LPST Number of Blows Figure 3.83.  Relationship between long-pin shear test number of blows and long-pin shear test torque value. 50 40 SPRT Torque, ft-lb 30 20 10 y = 0.7085x + 8.7120 R² = 0.7490 0 0 10 20 30 40 50 LPST Number of Blows Figure 3.84.  Relationship between long-pin shear test number of blows and short-pin raveling test torque value.

95   14 12 10 DPI, mm/blow 8 6 4 2 y = -0.1417x + 9.7401 R² = 0.5422 0 0 10 20 30 40 50 LPST Number of Blows Figure 3.85.  Relationship between long-pin shear test number of blows and DPI. 20 15 SPRT Number of Blows 10 5 y = 0.1150x - 0.1874 R² = 0.7855 0 0 20 40 60 80 100 120 140 160 180 LPST Torque, ft-lb Figure 3.86.  Relationship between long-pin shear test torque value and short-pin raveling test number of blows. 50 40 SPRT Torque, ft-lb 30 20 10 y = 0.2548x + 5.4441 R² = 0.7667 0 0 20 40 60 80 100 120 140 160 180 LPST Torque, ft-lb Figure 3.87.  Relationship between long-pin shear test torque value and short-pin raveling test torque value.

96 14 12 10 DPI, mm/blow 8 6 4 2 y = -0.0481x + 10.1612 R² = 0.4946 0 0 20 40 60 80 100 120 140 160 180 LPST Torque, ft-lb Figure 3.88.  Relationship between long-pin shear test torque value and DPI. 50 40 SPRT Torque, ft-lb 30 20 10 y = 2.0806x + 7.0908 R² = 0.8613 0 0 5 10 15 20 SPRT Number of Blows Figure 3.89.  Relationship between short-pin raveling test number of blows and short-pin raveling test torque value. 14 12 10 DPI, mm/blow 8 6 4 2 y = -0.4173x + 10.0762 R² = 0.6274 0 0 5 10 15 20 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- ches pleted by ensuring that the DCP handle was kept plumb and 12 in that testing was not started on a large piece of aggregate at the 4 inches 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 Shear or raveling fixture automated counters, which could reduce the number of operators needed. Operating the DCP by recording the dis- Figure 3.91.  Plan view of torque tance to a predetermined number of blows was considered as application. 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 torque rate, and the maximum value was always obtained base plate was kept parallel to the surface of the recycled layer within this distance. Figure 3.91 shows an illustration of while the shear and raveling fixtures were driven into the this process. recycled material. For both tests, the operator listened for a During initial stages of the field testing, there was concern change in the sound while driving the fixture to note when that the recycled material might be susceptible to raveling the base plate was touching the surface of the recycled layer. after completion of the field tests given that most of the The rate at which the torque was applied was kept constant projects were opened to traffic soon after the tests were com- from location to location by drawing a line (using a lumber pleted. The most destructive test devices included the DCP crayon on the recycled surface) 12 in. long from the center and the shear and raveling fixtures. Following each of these of the base plate. Perpendicular to this 12-in. line, and at tests, any disturbed material could be tamped back into the end farthest from the center of the base plate, a 4-in. place by the operator simply stepping on and compressing the line was drawn in the direction the torque wrench was disturbed material. Figure 3.92 shows the typical condition to be pulled by the operator. During testing, the operator of the pavement surface just after testing and again after the counted 4 seconds and, during this time, swung the arm of disturbed material had been tamped. Figure 3.93 shows the the torque wrench along the entire length of the 4-in. line. condition of the pavement surface 4 days after testing having This method allowed the operator to apply a consistent no deterioration caused by the testing. Figure 3.92.  Recycled material after a torque test (left) and after tamping (right).

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

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

100 Table 3.63.  DPI average, standard deviation, and coefficient of variation. DPI, mm/blow Standard Deviation Coefficient of Variation, % Test Cell/Material Average Single Multi- Single Multi- Operator Laboratory Operator 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 1.4 Single-Operator 1.2 Multi-Laboratory Standard Deviation, mm/blow Linear (Single-Operator) 1.0 Linear (Multi-Laboratory) R² = 0.67 0.8 0.6 0.4 R² = 0.27 0.2 0.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 DPI, mm/blow Figure 3.95.  Relationship between average DPI measurements and standard deviation. 25 Single-Operator Multi-Laboratory 20 Coefficient of Variation, % Linear (Single-Operator) Linear (Multi-Laboratory) R² = 0.56 15 10 R² = 0.04 5 0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 DPI, mm/blow Figure 3.96.  Relationship between average DPI measurements and coefficient of variation.

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

102 Table 3.66.  k-values for single-operator data consistency for LPST number of blows. Number of Blows Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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 3.67.  h-values for between-laboratory data consistency for LPST number of blows. Number of Blows Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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. by Laboratory 1, with a k-value of 1.69, the same observation blows and torque value, respectively, as calculated for each as with the case of number of blows measurements. For the test cell in accordance with ASTM C802. same reasons as stated previously, the data associated with To determine the form of the precision statements, the this observation were not removed from the analysis. relationship between the average LPST number of blows and To evaluate interactions among laboratories and test cells, the standard deviation and the COV for single-operator the LPST number of blows and torque value are shown in and multi-laboratory conditions is shown in Figures 3.99 Figures 3.97 and 3.98, respectively. The data are arranged and 3.100, respectively. The figures indicate that the stan- from least to greatest value. The trends from each laboratory dard deviation is also the appropriate basis for developing were similar for both the number of blows and torque value. the precision statements for LPST number of blows mea- From the results shown in Figures 3.97 and 3.98, no data were surements. That is because the standard deviation tends to excluded because of interactions. be relatively more independent of LPST number of blows Tables 3.70 and 3.71 show the single-operator and multi- measurements than the COV for both single-operator and laboratory standard deviation and COV for LPST number of multi-laboratory conditions. Thus, the pooled single-operator Table 3.68.  k-Values for single-operator data consistency for LPST torque values. Torque Value, ft-lb Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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 3.69.  h-values for between-laboratory data consistency for LPST torque values. Torque Value, ft-lb Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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.

103   50 45 Lab1 Lab2 Lab3 40 LPST Number of Blows 35 30 25 20 15 10 5 0 Cell 1LD Cell 1 Cell 4 Cell 5 Cell 7 Cell 3 Test Cell Figure 3.97.  Average LPST number of blows arranged from least to greatest. 180 Lab 1 Lab 2 Lab 3 160 140 LPST Torque, ft-lb 120 100 80 60 40 20 0 Cell 1LD Cell 5 Cell 4 Cell 1 Cell 7 Cell 3 Test Cell Figure 3.98.  Average LPST torque arranged from least to greatest. Table 3.70.  LPST number of blows average, standard deviation, and coefficient of variation. Number of Blows Standard Deviation Coefficient of Variation, % Test Cell/Material Average Single Multi- Single Multi- Operator Laboratory Operator 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

104 Table 3.71.  LPST torque value average, standard deviation, and coefficient of variation. Torque Value, ft-lb Standard Deviation Coefficient of Variation, % Test Cell/Material Average Single Multi- Single Multi- Operator Laboratory Operator 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 6 Single-Operator Multi-Laboratory 5 Linear (Single-Operator) Linear (Multi-Laboratory) Standard Deviation 4 R² = 0.16 3 2 R² = 0.16 1 0 0 10 20 30 40 50 LPST Number of Blows Figure 3.99.  Relationship between average LPST number of blows and standard deviation. 30 Single-Operator Multi-Laboratory 25 Linear (Single-Operator) Coefficient of Variation, % Linear (Multi-Laboratory) 20 15 R² = 0.29 10 R² = 0.24 5 0 0 10 20 30 40 50 LPST Number of Blows Figure 3.100.  Relationship between average LPST number of blows and coefficient of variation.

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

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

107   Table 3.74.  k-values for single-operator data consistency for SPRT number of blows. Number of Blows Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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 3.75.  h-values for between-laboratory data consistency for SPRT number of blows. Number of Blows Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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. The single-operator and multi-laboratory standard devia- deviations from Table 3.78 were used to develop the precision tion and COV for SPRT number of blows and torque value are statements for SPRT number of blows. presented in Tables 3.78 and 3.79, respectively, as calculated The relationship between average SPRT torque values and for each test cell in accordance with ASTM C802. the standard deviation and the COV for single-operator and To determine the form of the precision statements, the multi-laboratory conditions is presented in Figures 3.107 relationship between the average SPRT number of blows and and 3.108, respectively. The standard deviation tended to the standard deviation and the COV for single-operator increase with an increase in the SPRT torque value for both and multi-laboratory conditions is shown in Figures 3.105 single-operator and multi-laboratory conditions, and the and 3.106, respectively. The figures show that the COV is single-operator and multi-laboratory COV stayed relatively the appropriate basis for developing the precision statements constant with changes in the LPST torque value. Thus, the use for SPRT number of blows as the COV overall tended to be of a constant COV is appropriate for developing the precision relatively more independent than the standard deviation for statements for SPRT torque values. The pooled single-operator both single-operator and multi-laboratory conditions. Thus, and multi-laboratory COVs from Table 3.79 were used to the pooled single-operator and multi-laboratory standard develop the precision statements for SPRT torque values. Table 3.76.  k-values for single-operator data consistency for SPRT torque values. Torque Value, ft-lb Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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 3.77.  h-values for between-laboratory data consistency for SPRT torque values. Torque Value, ft-lb Laboratory Cell 1 Cell 1 Cell 3 Cell 4 Cell 5 Cell 7 FDR E-C FDR E-C LD CIR F-C CIR E-N CCPR E-N 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.

108 18 Lab1 Lab2 Lab3 16 14 SPRT Number of Blows 12 10 8 6 4 2 0 Cell 1LD Cell 4 Cell 1 Cell 5 Cell 7 Cell 3 Test Cell Figure 3.103.  Average SPRT number of blows arranged from least to greatest. 50 Lab1 Lab2 Lab3 45 40 SPRT Torque, ft-lb 35 30 25 20 15 10 5 0 Cell 1LD Cell 5 Cell 1 Cell 4 Cell 7 Cell 3 Test Cell Figure 3.104.  Average SPRT torque value arranged from least to greatest. Table 3.78.  SPRT number of blows average, standard deviation, and coefficient of variation. Number of Blows Standard Deviation Coefficient of Variation, % Test Cell/Material Average Single Multi- Single Multi- Operator Laboratory Operator 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

109   Table 3.79.  SPRT torque value average, standard deviation, and coefficient of variation. Torque Value, ft-lb Standard Deviation Coefficient of Variation, % Test Cell/Material Average Single Multi- Single Multi- Operator Laboratory Operator 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 4 Single-Operator Multi-Laboratory 3 Linear (Single-Operator) Linear (Multi-Laboratory) Standard Deviation 2 R² = 0.00 1 R² = 0.77 0 4 6 8 10 12 14 16 18 SPRT Torque, ft-lb Figure 3.105.  Relationship between average SPRT number of blows and standard deviation. 27 Single-Operator 24 Multi-Laboratory R² = 0.18 21 Linear (Single-Operator) Coefficient of Variation, % Linear (Multi-Laboratory) 18 15 12 9 6 R² = 0.06 3 0 4 6 8 10 12 14 16 18 SPRT Number of Blows Figure 3.106.  Relationship between average SPRT number of blows and coefficient of variation.

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

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

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

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