Experimental Study on a Prediction Model of the Shrinkage and Creep of Recycled Aggregate Concrete

Round 1

Reviewer 1 Report

The reviewed article titled: „Experimental study on prediction model of shrinkage and creep of recycled aggregate concreto”, in the manuscript were analyzed of the effects of recycled aggregate replacement rate and water-cement ratio on shrinkage and creep properties. The research conducted in the manuscript is interesting, but the article requires major revision, the observation is presented below:

A review of the literature to a large extent justifies taking up the research problem, but it is worth diversify and adding more international items.

In section 3. lists the names of 5 models from the literature. However, the itself models are not cited in the manuscript. There is no comment on the reason for choosing these models, although information is provided that there is many more formulas. There is also a lack of information on how they combine with the research presented in the manuscript. This issue requires a broader description at the manuscript.

Then in section 3.1. attempt to create their own predictive model. The section provides a series of equations, however, no reference to them in the text. The components of individual equations lack descriptions and information about units, e.g. manuscript, line 235 "V is volume; m is quality; p is apparent density. "

The models obtained in the manuscript should be confirmed by statistical analysis. Starting from the stage before creating the model. Present the applied methods of reaching the obtained model and information justifying the correctness of the statistical inference.

The inference path presented by the authors does not allow to verify the usefulness of the resulting model and thus the possibilities of its practical application. There is also no description of the limits of the applicability of the presented model.

Author Response

Response to Reviewer 1 Comments

 

Point 1: “A review of the literature to a large extent justifies taking up the research problem, but it is worth diversify and adding more international items.”

Response 1: Based on the comments of Reviewer 1, Nikola Tošic, R.V. Silva and Liu’s articles on shrinkage and creep are added to the introduction. Added content as shown follows and the rest of the changes could be found in the manuscript.

“Tošic et al. [20] explored the effect of recycled aggregates on the creep of concrete. It is pointed out that when using the creep prediction model of fib Model Code 2010 to predict the creep coefficient of RAC, the creep coefficient of RAC is underestimated relative to the performance of the model on the accompanying NAC. At the same time, compared with NAC, RAC shows a larger creep coefficient and the average increase in the coefficient of creep is 39% for RAC at full replacement rate. Silva et al. [21] studied the correction coefficient of concrete creep of different recycled aggregates and the prediction model suitable for recycled aggregate concrete. And compared with conventional concrete, the use of recycled aggregate to absorb part of the mixed water and cement slurry for coating produces a stronger ITZ, which can reduce the creep strain by up to 23%. Liu et al. [22] explored the calculation method of long-term deformation of recycled concrete beams based on creep adjustment coefficient. At the same time, the adjustment coefficient of mortar creep bond was proposed. Three typical ordinary concrete shrinkage and creep prediction models were modified and used for long-term deformation calculation of recycled concrete beams.”

 

Point 2: “In section 3. lists the names of 5 models from the literature. However, the itself models are not cited in the manuscript. There is no comment on the reason for choosing these models, although information is provided that there is many more formulas. There is also a lack of information on how they combine with the research presented in the manuscript. This issue requires a broader description at the manuscript.”

Response 2: There are many types of existing concrete shrinkage and creep prediction models, and the following are more influential and more applied.

International: CEB-FIP (1990) model recommended by the European Concrete Committee and the International Prestressed Concrete Association (CEB-FIP); the GL2000 model proposed by the American Concrete Association (ACI); the (ACI) 209R (1992) model recommended by the American Concrete Association (ACI) 209 Committee; British standard BS series model.

In China: The industrial standard of the People's Republic of China, "Design Specification for Highway Reinforced Concrete and Prestressed Concrete Bridges and Culverts", JTG D62-2004, used to predict the shrinkage and creep of concrete; "Concrete Structure Design Code" GB50010-2010 is used to calculate the prestress loss of concrete shrinkage prediction model. This is referred to as the GB50010 model.

Yang X.B. (‘Research on prediction model of concrete shrinkage and creep [D].’ Wuhan: Wuhan University, 2004) considered that among the three foreign newer prediction models (ACI) 209R (1992) model, CEB-FIP (1990) model, and GL2000 model, the overall performance of the GL2000 model is the best, followed by CEB- The FIP (1990) model, (ACI) 209R (1992) model performed the worst. At the same time, in the prediction model adopted by China's standard "Code for Design of Concrete Structures" GB50010-2010, the calculation of creep coefficient is consistent with the CEB-FIP (1990) model, but the prediction model of shrinkage strain is not the same. Therefore, based on the factor of recycled aggregate attached mortar, the model of this article is discussed in depth with five models.

 

Point 3: “Then in section 3.1. attempt to create their own predictive model. The section provides a series of equations, however, no reference to them in the text. The components of individual equations lack descriptions and information about units, e.g. manuscript, line 235 "V is volume; m is quality; ρ is apparent density. "

Response 3: According to the reviewer's suggestion, the units and information of the factors in the equation are added to the article. Added content as shown follows.

“In the equation: V is volume (m³); m is quality (kg); ρ is apparent density (kg/m³).”

 

Point 4: “The models obtained in the manuscript should be confirmed by statistical analysis. Starting from the stage before creating the model. Present the applied methods of reaching the obtained model and information justifying the correctness of the statistical inference.”

Response 4: Thanks to the comments made by the reviewers, the reasons for using the above model and the verification of its correctness mainly include the following aspects.

 The CEB-FIP (1990) model is one of the models recommended by the European Concrete Committee and the International Prestressed Concrete Association (CEB-FIP). At the same time, Marcelo Pedreira da Silva (The Behavior of Recycled Concrete through the Application of an Isotropic Damage Model, Scientific Research, Vol.5 No.3, September 2015) and other scholars also pointed out that it has better accuracy when studying recycled concrete models. The 209R (1992) model is an earlier model recommended by the American Concrete Association (ACI) 209 committee, which is cited and affirmed by scholars at home and abroad. It turns out that it is more correct and reasonable than some models. The GB50010 model is a prediction model of concrete shrinkage and creep for calculating prestress loss in China's "Concrete Structure Design Code" GB50010-2010. This model has been used as a model within the specification since 2010 and is one of several models widely recognized in the Chinese code. However, the adjustment coefficient proposed in this paper based on the attached mortar is based on these models, so its correctness has a good guarantee.

 

Point 5: “The inference path presented by the authors does not allow to verify the usefulness of the resulting model and thus the possibilities of its practical application. There is also no description of the limits of the applicability of the presented model.”

Response 5: The shrinkage and creep characteristics of recycled concrete are different from ordinary concrete mainly due to the adhesion of mortar to the recycled aggregate. And the usual way to prepare recycled concrete is to replace some or all of the natural aggregate with recycled aggregate. Therefore, starting from the factor of adhering mortar, considering its adjustment coefficient has important significance for the existing recycled concrete prediction model.

“Fathifazl et al. [18] considered that natural aggregate was replaced by mass or volume percentage when RAC was prepared by traditional methods. As compared with ordinary concrete, a certain amount of cement mortar is wrapped on the surface of the recycled aggregate. The content of natural aggregate in the RAC prepared by the conventional method is decreased and the content of the total mortar is increased. The elastic modulus of the component is decreased, and the shrinkage and the amount of creep are increased. It is assumed that the RCA is a two-phase material composed of residual cement mortar and natural aggregate, and the mortar in the RAC consists of residual cement mortar and new cement mortar on the aggregate surface. Based on the creep test results of RAC, the residual cement mortar content of RCA was taken as the research object to establish the creep prediction model of RAC.”

 The shrinkage of concrete is mainly caused by the hydration of cement mortar. Concrete is constrained by aggregates and its shrinkage is lower than that of pure cement. The recycled aggregate attached mortar will weaken the binding effect of the aggregate, so it is important to consider the adjustment coefficient of the attached mortar to the correction of the recycled concrete shrinkage model.

“Concrete is constrained by aggregates, so its dry shrinkage is lower than that of pure grout [33]. The shrinkage of concrete is mainly caused by hydration of cement mortar. The size of a few coarse aggregates is unstable. However, most of the coarse aggregates have the same size, large elastic modulus, high crush index and good restraint on concrete shrinkage. Therefore, the shrinkage of concrete depends on the amount and rigidity of the coarse aggregate in concrete”

The recycled concrete prediction model based on the adjustment coefficient of the attached mortar has good applicability from the lower replacement rate of the recycled aggregate to the full replacement rate. However, the model in this paper only considers the shrinkage and creep of the conventional recycled concrete. It cannot guarantee the accuracy of recycled concrete prepared from composite materials (such as fiber recycled concrete).

 

All the changes in the article are marked with bold and underline (E.g. changes).

Thanks to the reviewers and editors for reviewing the manuscript during the busy schedule and providing valuable comments.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors should improve the following aspects:

There are several works on this subject, some of which the authors failed to cite (see below); so, what is the innovation of the models proposed here relative to the ones in the literature? That innovation must be stated in the text of the paper; The physical properties of ALL aggregates must be provided; In Table 2, the water cement ratio should be identified as EFFECTIVE water cement ratio; The amount of additional water must be further explained; were the RCA oven-dried before being added to the mix and 3.83% is their total water absorption? What is the slump of the various mixes? In Figure 1, "contraction" should be changed to "shrinkage"; The literature review in terms of modelling the shrinkage and creep of recycled aggregate concrete is quite incomplete; among others, the works of Tosic et al. and Silva et al. are missing.

Author Response

Response to Reviewer 2 Comments

 

Point 1: “what is the innovation of the models proposed here relative to the ones in the literature? That innovation must be stated in the text of the paper.”

Response 1: Compared with the model used in the introduction and comparison, the prediction model obtained in this paper, considers the recycled aggregate attached mortar as a factor to propose the shrinkage and creep adjustment coefficient. Therefore, the innovation is that the model can more accurately reflect the special features of the recycled aggregate itself, that is, the characteristics of the attached mortar which is difficult to peel off compared to the natural aggregate. And it provides a simple calculation method for the promotion of recycled concrete under full replacement. Added content as shown follows and the rest of the changes could be found in the manuscript.

“The key difference between recycled aggregate and natural aggregate is the difference of the content of the attached mortar. In this paper, the contraction point is used to calculate the shrinkage strain and creep coefficient of the RAC shrinkage model, with the attached mortar increase coefficient method. The calculated value of the model takes into account the effect of the recycled mortar itself on the shrinkage and creep of the recycled concrete. The difference between the calculated value and the experimental value is small and can meet the accuracy requirements well.”

 

Point 2: “The physical properties of ALL aggregates must be provided”

Response 2: According to the reviewer's opinion, the physical properties of the aggregate were added in the article. Added content as shown follows and the rest of the changes could be found in the manuscript.

“Table 1. Physical properties of RCA and NCA.

Aggregate type	Apparent density /kg.m-3	Crushing index /%	Moisture content /%	Water absorption rate/%	Initial stone /%	Secondary aggregate /%	Mortar block /%	Impurity /%
Recycled coarse aggregate	2458	17.0	1.33	3.83	29.5	51.2	16.0	3.3
Natural coarse aggregate	2658	10.6	4.21	0.69	98.7	—		1.3

”

 

Point 3: “In Table 2, the water cement ratio should be identified as EFFECTIVE water cement ratio; The amount of additional water must be further explained. "

Response 3: According to the reviewer's suggestion, the water-cement ratio was modified to an effective water-cement ratio and the additional water intake was further explained. Added content as shown follows and the rest of the changes could be found in the manuscript.

 

“Table 2. Mix ratio of RAC and NAC.

Group number	r/%	Effective water-cement ratio	Material consumption /kg·m-3						Compressive strength /N/mm2
			cement	water	Natural coarse aggregate	Recycled coarse aggregate	Fine aggregate	Additional water
NAC-Ⅰ	0	0.527	370	195	1185	-	660	-	33.3
RAC-50-Ⅰ	50	0.527	370	195	592.5	592.5	660	22.69	31.6
RAC-100-Ⅰ	100	0.527	370	195	-	1185	660	45.38	32.4
NAC-Ⅱ	0	0.40	500	200	1086	-	611	-	35.1
RAC-50-Ⅱ	50	0.40	500	200	543	543	611	20.8	32.3
RAC-100-Ⅱ	100	0.40	500	200	-	1086	611	41.6	30.9

 

The additional water is calculated to take 3.83% of the weight of the recycled coarse aggregate under consideration of the water absorption rate of the recycled coarse aggregate.”

 

Point 4: “were the RCA oven-dried before being added to the mix and 3.83% is their total water absorption?”

Response 4: Since the surface of the recycled aggregate has a layer of adhering mortar which is difficult to peel off, and the mortar content is much higher than that of the natural aggregate, the water absorption rate of 3.83% refers to the total amount of the recycled coarse aggregate and the attached mortar.

The water absorption rate of recycled aggregates is measured by the test of recycled aggregates of the same batch of materials before pouring concrete.

 

Point 5: “What is the slump of the various mixes?”

Response 5: For the reviewer's opinion, the concrete slump recorded during the test was about 90mm.

 

Point 6: “In Figure 1, ‘contraction’ should be changed to ‘shrinkage’.”

Response 6: The "contraction" had been changed to "shrinkage" as recommended by the reviewer.

 

Point 7: “The literature review in terms of modelling the shrinkage and creep of recycled aggregate concrete is quite incomplete; among others, the works of Tosic et al. and Silva et al. are missing.”

Response 7: Based on the comments of Reviewer 2, Nikola Tošic and R.V. Silva's articles on shrinkage and creep are added to the introduction. Added content as shown follows and the rest of the changes could be found in the manuscript. 

“Tošic et al. [20] explored the effect of recycled aggregates on the creep of concrete. It is pointed out that when using the creep prediction model of fib Model Code 2010 to predict the creep coefficient of RAC, the creep coefficient of RAC is underestimated relative to the performance of the model on the accompanying NAC. At the same time, compared with NAC, RAC shows a larger creep coefficient and the average increase in the coefficient of creep is 39% for RAC at full replacement rate. Silva et al. [21] studied the correction coefficient of concrete creep of different recycled aggregates and the prediction model suitable for recycled aggregate concrete. And compared with conventional concrete, the use of recycled aggregate to absorb part of the mixed water and cement slurry for coating produces a stronger ITZ, which can reduce the creep strain by up to 23%.”

 

All the changes in the article are marked with bold and underline (E.g. changes).

Thanks to the reviewers and editors for reviewing the manuscript during the busy schedule and providing valuable comments.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thank you for your replies to the article review " Experimental study on prediction model of shrinkage and creep of recycled aggregate concrete", but in my opinion the article needs further revision.

It is good that the authors have completed the units in line 235 but this is only an example. Units and descriptions of designations in patterns should be completed in the whole text, including diagrams and tables.

Following the authors: "Concrete Structure Design Code" GB50010-2010 is used to calculate the prestress loss of concrete shrinkage prediction model. This is referred to as the GB50010 model. Yang X.B. (‘Research on prediction model of concrete shrinkage and creep [D].’ Wuhan: Wuhan University, 2004) considered that among the three foreign newer prediction models (ACI) 209R (1992) model, CEB-FIP (1990) model, and GL2000 model, the overall performance of the GL2000 model is the best, followed by CEB- The FIP (1990) model, (ACI) 209R (1992) model performed the worst. At the same time, in the prediction model adopted by China's standard "Code for Design of Concrete Structures".

The authors use the terms "worst" and "best" for no statement is given a specific numerical value. How much better, how much worse?  For example, in percentage terms. He proposes to make a comparative table with the results, using particular models.

Unfortunately, the authors did not comment on the statistical analysis of the quality of the tests and the model used by the authors. The reliability of the presented research is very important for the possibility of future application of the research.

Unfortunately, the article has been omitted.

Author Response

Response to Reviewer 1 Comments

 

Point 1: “It is good that the authors have completed the units in line 235 but this is only an example. Units and descriptions of designations in patterns should be completed in the whole text, including diagrams and tables.”

Response 1: Based on the reviewers' comments on the need to add units to each chart, the factors in the chart are added to the comments and units. The changes are as follows, and the rest of the changes (identified as bold and underlined) can be found in the manuscript.

“Note:  indicates shrinkage strain and is dimensionless; indicates the slope and is dimensionless;  indicates the concrete age at the start of drying, t indicates the age of the concrete.

Note:  indicates shrinkage strain and is dimensionless;  indicates the slope and is dimensionless;  indicates the loading age of concrete, t indicates the age of concrete.”

 

Point 2: “The authors use the terms "worst" and "best" for no statement is given a specific numerical value. How much better, how much worse?  For example, in percentage terms. He proposes to make a comparative table with the results, using particular models.

Unfortunately, the authors did not comment on the statistical analysis of the quality of the tests and the model used by the authors. The reliability of the presented research is very important for the possibility of future application of the research.”

Response 2: Based on the experimental results of the computer database, the average coefficient of variation of the CEB-FIP (1990) model creep function is estimated to be approximately 20% (1990 CEB-FIP model specification (concrete structure) [S], School of Building Research, 1991, 12: 57-70). The ACI209R model has characteristics that do not distinguish between elastic deformation and plastic deformation and does not consider the influence of concrete strength when calculating creep. The ACI209R model is in poor agreement with the experimental data and often underestimates the effects of shrinkage and creep effects (David, W. Mokarem. Development of concrete shrinkage performance specification [D]. Faculty of the Virginia Polytechnic Institute and State University, 2002, 5), And Du et al. (Du, C.C. Study on the Influence of Construction Load on Vertical Deformation of High-rise Building Structure [D]. Harbin Institute of Technology, 2013, 6) found that the ACI209R model compared with the experimental results, the average coefficient of variation can reach 40%. Gardner et al. (N. J. Gardner, Comparison of Prediction Provisions for Drying Shrinkage and Creep of Normal Strength Concretes. Canadian Journal of Civil Engineering, 2011, 31(5):767-775) have experimentally compared that the GL2000 model has better accuracy than the ACI209 and 1990 models. “Using all available input information GL 2000 and B3 are significantly better able to predict shrinkage with normalised RMS errors (coefficients of variation) of 19-20%. Using all available input information GL 2000 is the best for compliance followed by B3 and CEB MC 1990-99. Using only the average concrete strength as input information, GL 2000 is better than the other methods for predicting shrinkage with a normalised RMS error (coefficient of variation) of 25%, but for compliance the normalised RMS error increases to 26% which is similar to that of B3.” Therefore, with reference to the research of many scholars, we consider the pros and cons of the model as the content of the answer for reviewer. And the reliability of the attached mortar adjustment coefficient based on the above model is also considered to be applied well.

 

 

All the changes in the article are marked with bold and underline (E.g. changes).

Thanks to the reviewers and editors for reviewing the manuscript during the busy schedule and providing valuable comments.

Author Response File: Author Response.pdf

Reviewer 2 Report

All comments from the previous review have been adequately addressed except one:

- The slump of all the mixes should be provided in the manuscript.

Author Response

Response to Reviewer 2 Comments

 

Point 1: “The slump of all the mixes should be provided in the manuscript.”

Response 1: For the reviewer's opinion, we added the slump indicator for the concrete mixture to the manuscript. The modified content is added at the 1.2. Mix proportion design. The changes are as follows.

“The concrete slump of each component during mixing is about 90mm.”

 

 

All the changes in the article are marked with bold and underline (E.g. changes).

Thanks to the reviewers and editors for reviewing the manuscript during the busy schedule and providing valuable comments.