Full-process temperature control and crack prevention of roller compacted concrete in water conservancy projects

1 Introduction

Roller compacted concrete (RCC) has become the mainstream dam-building material for large-scale hydraulic structures such as gravity dams, cofferdams, and spillways, due to its high construction efficiency, low cement content and excellent impermeability and durability (Nasiri et al., 2024; Chen and Chen, 2023). However, during the hardening process, RCC is prone to significant temperature gradients and autogenous volume deformations, resulting from concentrated heat release during cement hydration, low thermal conductivity, and external restraints, which can induce temperature cracks. Such cracks not only compromise the integrity of structures but may also become seepage channels, endangering the long-term safe operation of projects (Sok et al., 2022). Therefore, how to optimize the full-process temperature control of RCC—covering raw material proportioning, mixing and transportation, layer-by-layer pouring, and subsequent curing—to effectively prevent the formation of cracks has become a critical technical issue in modern water conservancy engineering construction (Jin et al., 2022). Traditional temperature control measures have mostly focused on post-cooling, lacking systematic coupling of the intrinsic thermal properties of materials with the dynamic process of construction (Liu et al., 2021). With the development of intelligent sensing, numerical simulation, and materials science, the concept of full-process, multi-scale, proactive temperature control has gradually emerged, such as reducing adiabatic temperature rise at the source through mix proportion design, regulating thermal boundary conditions via construction technologies, and adjusting curing strategies based on real-time monitoring and feedback (Jayantha et al., 2024). Sun (2025) optimized the temperature control scheme for high RCC dams in northern Xinjiang, China, using computational programs based on the ANSYS platform. Ünverdi et al. (2025) comprehensively examined the effects of different fiber types on improving the high-temperature performance of RCC. They found that high temperatures cause significant changes in the physical and mechanical properties of concrete, and steel fibers and environmentally friendly sustainable scrap steel fibers can reduce strength loss by limiting the expansion of micro-cracks within the concrete matrix. Sun et al. (2024) systematically investigated the effects of initial damage and freeze-thaw cycles on the physical and mechanical properties of RCC through a series of physical tests, including freeze-thaw cycle tests, pore structure tests, and uniaxial compression tests. Peng et al. (2025) adopted a new hydration model to more accurately reflect the changes in temperature and hydration degree during the early hydration stage. Through simulation, they found that the improved hydration model accelerated the hydration rate and advanced the time of crack appearance and penetration. Al-Khafaji et al. (2024) explored the influence of using carbon fiber-reinforced polymer (CFRP) plates on the flexural performance of self-compacting concrete beams under repeated loads. They discovered that a limited number of repeated loads had a relatively small impact on the flexural performance of the beams, resulting in a maximum decrease of only 11% in the ultimate load. Chen et al. (2023) proposed a new crack identification method suitable for underwater concrete structures. They found that the distribution characteristics of ψ along the monitoring line can accurately locate the cracks but cannot quantify the crack width. In this paper, the full-process temperature control optimization and crack prevention model test of RCC was conducted. A reference group, five single-factor optimization groups and one full-process integrated optimization group were set up to compare and analyze the effects of measures such as mix proportion, mixing temperature, pouring technique, embedded cooling water pipes, and post-curing on temperature field, stress field, and crack resistance. The innovations of this paper lie in the following aspects: constructing a phased incremental control design; using a reduced-scale model to equivalently restore the laws of the actual dam; and adopting multi-index synchronous continuous monitoring of crack resistance performance. The limitation of this paper is that in the phased incremental control experiment, one phase was treated as an overall factor for comparing differences, without further analyzing the differences among different factors within that phase.

2 Case study

2.1 Experimental materials

Experimental materials included ordinary silicate cement (PO 42.5, Henan Yong’an Bridge and Tunnel Building Materials Co., Ltd., China), low-heat cement (PLH, Beijing Haiyan Xingye Concrete Admixture Sales Co., Ltd., China), fly ash (Lingshou County Shuntian Mineral Products Processing Plant, China), fine aggregate (granite sand, 0–4 mm particle size), and coarse aggregate (common gravel, 5–8 mm particle size).

2.2 Scheme setup

The main cause of cracks in RCC is that the temperature stress exceeds the tensile strength of the concrete. The formation process can be divided into the temperature rise period (0–3 days), the temperature drop period (3–28 days), and the long-term stabilization period (beyond 28 days) (Samanasa et al., 2024). During these periods, heat generation from cement hydration induces internal temperature variations, creating a temperature differential between the interior and exterior of the concrete mass. This differential generates both thermal tensile stress and shrinkage-induced tensile stress. When combined with fluctuations in ambient temperature, these stresses can initiate cracking (Kumar et al., 2025). Traditional concrete temperature control measures usually involve surface watering for cooling and curing after pouring, but provide little control over the internal temperature of the concrete. To prevent RCC cracking, a full-process temperature control strategy has been proposed (Boussetta and Khay, 2025). Based on the heat generation mechanism of RCC and in combination with the RCC pouring process, this strategy can be divided into four stages: mix proportioning, mixing, pouring, compaction, and post-curing. The mix proportioning stage aims to reduce the heat generation of RCC at the source; the mixing stage controls the heat generated during the mixing and stirring of RCC; the pouring and compaction stage reduces the heat generated through the pouring method; and the post-curing stage controls heat dissipation (He et al., 2024).

2.2.1 Analysis of similarity of a reduced-scale gravity dam model

A 1:40 geometrically scaled block model was used in the experiment to simulate the large-volume RCC structure of an actual gravity dam. To ensure that the temperature field, stress field, and crack evolution law of the model were consistent with those of the prototype, similarity criteria were derived from six aspects: geometry, hydration heat, temperature, cooling aging, stress, and cracking behavior, so that the experimental results can equivalently restore the temperature control and anti-cracking laws of the prototype project.

In terms of geometric similarity, the length, width, and height of the scaled block model maintained a unified scaling ratio. The pouring layer thickness, the diameter and spacing of cooling water pipes, and the thickness of the insulation layer were simultaneously scaled down by a ratio of 1:40 to ensure that the internal heat dissipation channels and the ratio of the boundary heat transfer area were completely consistent with those of the prototype.

In terms of the similarity of hydration heat sources, the adiabatic temperature rise of RCC was controlled by the total amount of cementitious material hydration. The heat conservation similarity criterion was adopted, i.e., the total hydration heat release per unit volume of the model was equal to that of the prototype. In the test, the same cement, fly-ash, and aggregate system as the prototype was used, and the water-binder ratio, total amount of cementitious materials, and mineral admixture content remained unchanged. There was no scale reduction in the hydration heat per unit volume, and only the heat conduction time scaled with the geometric size. The theoretical peak temperature rise per unit volume of the model and the prototype was the same, and there were only scale differences in the duration of temperature rise and fall.

In terms of temperature field similarity, temperature was a dimensionless thermodynamic quantity, and the temperature similarity ratio was 1. The maximum central temperature, the temperature difference between the inside and outside, and the allowable temperature difference limit of the model were completely equivalent to those of the prototype. The temperature, humidity, wind speed, and the diurnal temperature fluctuation cycle in the environmental chamber strictly replicated the on-site summer working conditions. The boundary convective heat transfer coefficient h was the same as the value of the prototype. Only the heat conduction time was scaled according to the scale. The temperature difference measured by the model directly corresponded to the real inside-outside temperature difference of the prototype, and no temperature correction was required.

In terms of cooling time similarity, the Fourier similarity number of heat conduction was equal. The mix proportion of the model was consistent with that of the prototype, leading to a same thermal diffusivity. Thus, the temperature evolution of the prototype could be reproduced in a shorter time in the scaled model, which significantly reduces the cooling and curing time of the model.

In terms of stress field similarity, during the linear elastic stage, the stress, elastic modulus, Poisson’s ratio, and linear expansion coefficient were completely consistent with those of the prototype concrete. The measured tensile stress value of the model could directly represent the real stress level of the prototype under the same temperature difference, and the stress reserve coefficient had engineering comparison and reference value.

In terms of the similarity of crack evolution behavior, the cracking judgment criterion was that surface or internal cracks occur when the tensile stress due to temperature exceeds the tensile strength of concrete. The test used concrete from the same batch, and there was no scale distortion in tensile strength, stress, and temperature difference. Therefore, the critical temperature difference for cracking, the temperature history corresponding to the first cracking, and the crack propagation driving mechanism were completely consistent with those of the prototype.

2.2.2 Settings of comparison schemes

To verify the effect of the full-process temperature control optimization strategy on preventing RCC cracking, this paper adopts a phased incremental optimization strategy, in which temperature control measures are progressively introduced to establish comparative schemes. The corresponding strategy parameters are shown in Table 1. G0 is the reference condition for comparison, representing the traditional RCC pouring method. G1 is a material optimization group, in which only the mix proportion is modified: ordinary cement is replaced with low-heat cement, the fly ash content is increased to 45%, and the water-to-binder ratio is adjusted to 0.50, while all other parameters remain consistent with G0. G2 is the mixing temperature control group, in which only the temperature during the mixing process is regulated. During the mixing process, the temperature of the cement was controlled at 12 °C through pre-cooling of aggregates (Pulecio-Díaz et al., 2024) and the addition of ice, while all other parameters remained identical to those of G0. G3 is the construction technology group, in which only the pouring method is optimized: the pouring layer thickness is reduced to 0.050 m, and the surface is covered with a 2-cm-thick insulation blanket, with the remaining parameter configuration kept the same as that of G0. G4 is the cooling system group, in which only water cooling is implemented within the concrete. During the pouring process, HDPE cooling pipes with an i`nner diameter of 25 mm are embedded at spacings of 37.5 mm, through which cooling water at 10 °C is circulated at a flow rate of 1.5 m³/h for a duration of 14 days, while all other parameters remain the same as those of G0. G5 is the curing reinforcement group, which only enhances the post-curing: the concrete surface is covered with a moisture-retaining film and subjected to water sprinkling for 28 days. G6 is the full-process optimization group, which integrates all the optimization measures from G1 to G5. Five specimens were prepared for each of the above sample groups. The mix proportions and the corresponding thermal and mechanical characteristic parameters of the different groups are presented in Table 2.

TABLE 1. Comparison scheme settings.

TABLE 2. The concrete mix proportions and their thermal and mechanical characteristic parameters of different groups.

2.3 Experimental procedures

The process of conducting experiments on the comparative working condition schemes, G0-G6, is shown in Figure 1.

First was the preparation stage, during which the environmental chamber was calibrated and a reduced-scale mold for typical gravity dam block was made. The parameters of the environmental chamber were set as follows: air temperature at 30 °C, relative humidity at 60%, wind speed at 2.0 m/s, a diurnal cycle comprising 12 h of high temperature and 12 h of low temperature, so as to simulate summer conditions.

The second stage was the pouring stage. The first day of this stage was denoted as day 0. Seven groups of specimens were poured on the same day, and the temperature of casting concrete was strictly controlled in accordance with the parameters specified for each respective scheme. During pouring, each specimen was placed in three layers, with an interval of 5 days between successive layers; moreover, the same layer of all seven groups was poured on the same day. The concrete was vibrated and compacted using a small vibratory tamper (Zhang et al., 2020).

The third stage was the curing and monitoring stage, which lasted from day 0 to day 40. The seven groups of specimens were poured, insulated, and watered according to their respective schemes. At the same time, T-type thermocouple sensors were embedded in the specimens during the pouring process to collect temperature variations. There was one thermocouple at the center of the specimen, four on the surface, and four between the layers. After the specimen was cast, vibrating wire strain gauges were used to measure stress variations within the specimen. Three strain gauges were arranged in the horizontal and vertical directions on the specimen surface.

The second and third steps were carried out in the environmental chamber throughout.

2.4 Evaluation indicators

Evaluation indicators for the specimens under different parameter configuration strategies included temperature control effect, crack resistance performance, and structural response. The temperature control effect included the maximum center temperature, the maximum internal and external temperature difference, and the cooling rate (measured at days 3–10).

Crack resistance included first crack time, total crack area, and maximum crack width. The structural response was maximum tensile strain. The steps for collecting and quantifying the surface crack images of the test specimens are as follows.

• Camera hardware and shooting parameter settings: An industrial high-definition CMOS camera was paired with a fixed-focus macro lens. A fixed vertical support was used to ensure that the shooting position always maintained a fixed object distance of 500 mm from the specimen surface. For the shooting environment, a non-stroboscopic light-emitting diode (LED) surface light source was uniformly adopted, arranged at a 45° side light to weaken reflections and enhance the contrast between cracks and the concrete substrate. The fixed camera parameters are as follows: the resolution was 4,000 × 3,000 pixels, ISO = 100, the shutter speed was 1/125 s, and the aperture was F8. Each shot fully covered the entire exposed observation surface of the specimen. The shooting time was fixed at the same time every day to eliminate measurement errors caused by the day-night difference in illumination.
• Image scale calibration: Standard 10 mm stainless steel scale rulers were placed at the four corners of each captured image as calibration references; an image calibration tool was used to extract the pixel length of the ruler to calculate the actual physical size of a single pixel.
• Image crack recognition: First, the color captured images were converted into grayscale images. Then, binary threshold segmentation was employed to divide the dark crack area and the bright concrete matrix area. After that, morphological filtering such as erosion and dilation operations was performed. Finally, the binary image was traversed to extract all closed crack contours.
• Calculation of crack width and area: 20 sets of vertical cross-sections were evenly taken along the normal direction of the extracted crack contour. The maximum pixel distance of the cross-sections was taken as the pixel width of the crack, and then the actual crack width was obtained by using the scale conversion coefficient. All the pixel points within the closed contour of a single crack were accumulated and counted. The total pixel area was multiplied by the actual area of a single pixel to obtain the real area of a single crack. The total crack area of the observation surface was obtained by accumulating the areas of all independent cracks on the specimen surface.

The structural responses include the maximum tensile strain and the stress reserve coefficient. The stress reserve coefficient characterizes the safety margin of the concrete’s tensile strength and reflects the relative relationship between the maximum temperature-induced tensile stress of the structure and the instantaneous tensile strength of the concrete at the same period. When the coefficient is greater than 1, it means that the tensile strength of the concrete can cover the temperature-induced tensile stress, and the structure will not develop temperature cracks. When the coefficient is less than 1, the temperature-induced tensile stress exceeds the tensile limit of the material, and there is a risk of cracking. The relevant formula is shown in Equation 1:where K f is the stress reserve coefficient, f t τ is the axial tensile strength of the concrete at the τ-th day after pouring, and σ max τ is the maximum temperature-induced tensile stress inside the concrete at the τ-th day after pouring.

K f = f t τ σ max τ (1)

2.5 Experimental results

Temperature variations in the different groups of specimens over the 0- to 40-day period were collected via thermocouples pre-embedded in the concrete, in order to evaluate the temperature control effect. The central temperature, surface temperature, and internal and external temperature difference change curves of different groups of specimens are shown in Figures 2–4. Over time, the central temperatures of all seven groups of specimens showed a trend of rising first and then decreasing, peaking on the third day before declining gradually. Day 0–3 was the temperature raising period, days 4–28 was the temperature drop period, and days 29–40 was the stable period. The temperature control effects of the different groups under their respective strategies are shown in Table 3. G0, representing the traditional RCC pouring method, exhibited the highest central temperature, the maximum internal and external temperature difference, and the highest cooling rate. G1-G6 all adopted their respective temperature control strategies, under which both the peak central temperature and the maximum internal and maximum temperature difference were reduced. Among them, the specimen adopting the full-process temperature control optimization strategy, demonstrated the most favorable temperature control performance, characterized by not only a small internal and external temperature difference but also a low cooling rate.

TABLE 3. Temperature control effects of different groups of specimens.

The crack resistance of the concrete specimens was evaluated by monitoring the surface cracks within 0–40 days using cameras, and the results are shown in Table 4. G0, which employed the traditional pouring method, exhibited the earliest cracking, with a total crack area of 128.5 mm² and a maximum crack width of 0.32 mm. For G1-G6, the adoption of temperature control strategies resulted in a delayed cracking occurrence time. G6 had the best performance, with no cracking within the 40 days.

TABLE 4. Crack resistance of specimens in different groups.

The structural stress variations of the concrete specimens collected, collected using vibrating wire strain gauges over 0–40 days, are summarized in Table 5. The evolution curves of the maximum tensile stress of the different groups are presented in Figure 5. G0, which adopted the traditional pouring method, exhibited the highest maximum tensile stress, whereas G6, employing the full-process temperature control strategy, showed the lowest maximum tensile stress. Based on the cracking observations of the seven groups, it was found that only G6 with a stress reserve coefficient of 1.43 remained uncracked, while the stress reserve coefficients of all other cracked specimens were below 1.

TABLE 5. Structural stress of different groups of specimens.

3 Discussion

RCC, owing to its outstanding advantages such as low heat of hydration, fast construction progress, and favorable cost-effectiveness, has become a core material for dam construction in large-scale hydraulic structures. However, due to the concentrated release of hydration heat within the massive concrete mass and limited heat dissipation capacity, significant internal and external temperature differences and temperature stresses are very likely to form during the hardening process. Once these stresses exceed the tensile strength of the concrete, temperature cracks may occur, directly affecting the integrity, impermeability, and long-term operation safety of projects. Traditional temperature control measures often rely on passive measures such as watering and curing in the later stage, lacking systematic optimization of the entire process involving materials, mixing, pouring, cooling, and curing, resulting in limited effectiveness. Therefore, this paper adopted a combined approach of single-factor progressive optimization and full-process integrated optimization to systematically compare the variations in temperature, stress, and crack under different temperature control schemes, thereby providing a scientific basis for temperature control and crack prevention in RCC water conservancy projects. This paper focuses on the full-process temperature control optimization and crack prevention of RCC. Seven sets of comparative conditions were designed to conduct single-factor optimization and full-process integrated optimization across five dimensions: mix proportion, mixing temperature, pouring technique, internal cooling, and post-curing. Then, the effectiveness of the temperature control measures was evaluated through the indicators of temperature, stress, and crack.

An analysis is conducted from the perspectives of hydration heat generation and the heat suppression mechanisms of various measures. The root cause of the internal temperature rise in RCC is the hydration exothermic reaction of Portland cement and fly ash. During the temperature rising period (Day 0–3), the hydration reaction rate is the fastest, and a large amount of heat is released per unit volume in a short time. Limited by the low thermal conductivity of concrete, the heat accumulates inside the structure to form a high-temperature core area, which is the fundamental inducement for the subsequent temperature gradient and tensile stress. There are essential differences in the suppression paths of hydration heat among different optimization measures: G1 adopts the scheme of low-heat cement +45% high-volume fly ash, which is a source heat suppression mechanism. Ordinary P・O cement has a fast hydration rate of tricalcium silicate and dicalcium silicate and a high exothermic peak. The mineral composition of low-heat P・LH cement is optimized, resulting in a significant slowdown in the early hydration rate and a remarkable reduction in the exothermic peak. As an active admixture, fly ash can replace a large amount of cement, greatly reducing the total heat release of the cementitious system. At the same time, the secondary hydration of fly ash is delayed until the temperature drop stage, dispersing the exothermic peak and avoiding concentrated heat generation at day 0–3. In the test, the peak central temperature of G1 was 9.2 °C lower than that of G0, directly reducing the total amount of heat sources. G2 aggregate pre-cooling with ice mixing does not change the total heat release of cementitious material hydration. It is an initial temperature regulation method and cannot reduce the total hydration heat. However, it can directly lower the initial temperature of concrete when it is placed into the formwork, reduce the initial temperature difference between the structure and the environment, and weaken the temperature gradient accumulation in the temperature rising period. G3 thin-layer pouring and G4 pre-buried cooling water pipes are heat dissipation enhancement measures. They do not change the total amount of heat generated by hydration but only accelerate the outward transfer of internal heat by expanding the heat dissipation area and adding internal heat dissipation channels, thus reducing the peak value of heat accumulation. G5 extended curing only regulates the surface heat dissipation rate and has no inhibitory effect on the total amount of hydration heat generation. It only slows down the temperature drop rate in the temperature drop period. Therefore, using it alone has the weakest heat suppression effect. The full-process integrated group G6 combines mechanisms of heat suppression at the source, initial temperature reduction, and simultaneous internal and external heat dissipation. It restricts the hydration temperature rise from the whole chain of heat generation, storage, and dissipation. The peak central temperature was only 41.9 °C, which was 15.4 °C lower than that of the reference group G0. This group significantly weakens the generation basis of temperature stress at the root.

An analysis is performed based on the formation law of temperature gradients and the regulatory differences among different groups. The internal and external temperature difference, i.e., the temperature gradient between the high-temperature area at the center of the structure and the low-temperature area on the surface, is the direct cause of the tensile stress on the surface. In the temperature rising stage, the interior heats up rapidly, while the surface temperature is relatively low due to convective heat dissipation to the environment, forming a “hot inside and cold outside” positive gradient. When entering the temperature drop period (day 3–28), the interior dissipates heat and cools down slowly, while the surface cools down rapidly due to temperature fluctuations in the air. The temperature gradient persists and induces continuous tensile stress.

The test results showed that the maximum internal and external temperature differences in each group all occurred at the peak temperature rise moment on the third day, and the magnitude of the temperature gradient directly determines the subsequent stress level. The reference group G0 had no temperature control measures. During the temperature rising period, a large amount of heat accumulated, and the maximum temperature difference reached 23.7 °C. The huge temperature gradient caused the surface concrete to be constrained by the internal expansion, generating high-intensity tensile stress. Group G1 reduced the upper limit of the temperature gradient by lowering the peak value of hydration heat. Group G2 lowered the initial temperature and simultaneously weakened the temperature difference accumulation during the temperature rising stage. Group G3 adopted thin-layer pouring to increase the inter-layer heat dissipation area, balanced the internal temperature distribution, and effectively reduced the steady-state temperature gradient. Therefore, this group had a prominent advantage in temperature difference control. Group G4 used water-cooling to quickly remove the internal heat, but the internal temperature dropped too fast in a short time, which was likely to form a short-term secondary temperature gradient. When used alone, its effect on temperature difference control was weaker than that of thin-layer pouring. Group G5 only provided surface insulation, which could only slightly slow down the surface cooling rate and was unable to relieve the core temperature difference caused by internal heat storage during the temperature rising stage. Therefore, the reduction of the temperature gradient was minimal.

G6 integrated all measures. It not only reduced the total heat generation at the source but also coordinated and balanced the temperature distribution across the whole area through pre-cooling, thin-layer pouring, internal water circulation, and surface insulation. The maximum internal and external temperature difference was only 15.1 °C, and the gradient drop exceeded 36%. It eliminated the conditions of extreme temperature differences in terms of spatial temperature distribution and significantly reduced the driving force for surface tension.

Analyzing from the evolution of tensile stress, the temperature-induced tensile stress is generated by the deformation constraint caused by the temperature gradient, and its evolution is divided into three periods. 1. Temperature rising period (day 0–3): The interior is heated and expands. The surface layer restricts the internal expansion, resulting in surface tensile stress. The stress rises synchronously with the temperature difference. 2. Temperature drop period (day 4–28): The whole structure contracts. The cooling and contraction rate of the surface layer is higher than that of the interior. Superimposed with the residual constraint generated by the early expansion, the tensile stress continues to accumulate and reaches its peak on the 7th day. 3. Stable period (day 29-): The temperature across the field approaches the ambient temperature, the deformation is coordinated, and the tensile stress decays slowly. In the experiment, the peak tensile stress of all groups lagged behind the peak temperature by 4 days, which proves that the cooling shrinkage is the core stage for controlling cracking. The paths for each group to suppress tensile stress are different. Group G1 reduces the hydration temperature rise and narrows the temperature gradient, thereby reducing the stress amplitude from the source of stress generation. Group G2 lowered the pouring temperature, reduced the total amount of overall deformation, and simultaneously weakened the stress growth during both the temperature rising and drop stages. Group G3 conducted thin-layer pouring to homogenize the temperature field, weakened the deformation difference between the inside and outside, and effectively reduced the steady-state tensile stress. Although Group G4 used water circulation for rapid cooling to reduce the peak temperature, the rapid shrinkage induced secondary stress, and the reduction of peak stress was limited when used alone. Group G5 only delayed the surface shrinkage rate and had no improvement on the residual constraints generated during the early-stage heating, suggesting the worst stress suppression effect. Group G6 used multiple measures to synergistically constrain the temperature-deformation difference. The maximum tensile stress was only 0.98 MPa, and the stress reserve coefficient was 1.43, indicating sufficient tensile safety margin. No micro-cracks were observed within the 40-day observation period.

A single-factor optimization group can only address the shortcoming of a single link in temperature control and has obvious limitations. G1 only reduced the heat generated by hydration but had no heat dissipation regulation, and a relatively large shrinkage stress was formed during the temperature drop stage. G2 only lowered the initial temperature and cannot suppress the heat storage due to later hydration. G3 only optimized heat dissipation without reducing the total amount of hydration heat. G4 was prone to generate secondary temperature differences when water was circulated. G5 only provided a late-stage remedy and cannot control the heat source and temperature gradient in the early stage. The optimization effects of various single measures are limited, and they cannot simultaneously meet the crack control conditions of low temperature rise, small temperature gradient, slow cooling, and low stress. There were significant synergistic gains in the full-process integrated optimization G6: low-heat cementitious materials reduced the total heat production (heat source control) + pre-cooling of aggregates lowered the initial temperature rise (initial temperature control) + thin-layer pouring balanced the internal temperature (spatial gradient balance) + HDPE water-cooling continuously removed the residual heat (active heat dissipation) + long-term heat and moisture preservation slowed down the surface shrinkage (cooling rate control). These five measures complement each other’s shortcomings, forming a closed-loop temperature control system, breaking through the performance ceiling of single measures, and achieving the optimal effect of no cracking in 40 days and a significant reduction in tensile stress. This fully demonstrates that the RCC dam must adopt the full-process integrated temperature control concept, and relying solely on local and post-event maintenance measures cannot meet the long-term anti-cracking safety requirements.

The material mix ratio, construction parameters, water-passing curing scheme, and temperature control threshold obtained from this 1:40 scale model can provide references for the construction of the actual RCC gravity dam. Combining the similarity criteria and on-site engineering constraints, the process is divided into five major links: material, pouring, cooling, monitoring, and zoning and grading control.

• Engineering transformation of raw materials and mix ratio: low-heat Portland cement is mixed with 45% fly ash, and the water-binder ratio is controlled at 0.50. The on-site engineering team fine-tunes the water consumption according to the moisture content of local aggregates. During high-temperature construction periods, an aggregate air-cooling and ice-adding mixing system outside the warehouse should be used to stably control the on-site concrete discharge temperature within 20 °C, replicating the pre-cooling and temperature control effect of G2. At the same time, parallel tests on the adiabatic temperature rise, elastic modulus, and linear expansion coefficient of concrete can be performed upon arrival at the site to verify the consistency of the thermal and mechanical parameters with those of indoor test specimens. If the heat conduction and heat release performance of local aggregates deviate, the fly ash content can be slightly adjusted to adapt to on-site conditions.
• Layered pouring process: For the strong constraint areas of the dam body (the contact surfaces of the foundation and the bank slopes), thin-layer rolling within 50 cm is uniformly adopted. The inter-layer intermittent period is controlled at about 5 days to avoid large-area heat accumulation in thick pouring blocks. The pouring block surface is fully covered with a 2-cm flame-retardant thermal insulation quilt. In case of strong wind or cold snap, a double-layer insulation layer is added to replicate the technological advantages of the balanced temperature gradient of G3. During the high-temperature season, pouring is preferably carried out during the low-temperature period at night, and the pouring block surface is sprayed with water for moisture retention simultaneously to reduce the sudden surface cooling caused by environmental convection.
• Standardized implementation of pre-buried water-cooling: Cooling pipe networks can be installed across the entire pouring blocks in the strong constraint areas of the dam body. A phased gradient water-passing system can be adopted. In high tensile stress areas such as corridors and notches, the water pipes are arranged more densely to further reduce the local temperature gradient.
• On-site intelligent monitoring and threshold control: Thermocouples and strain monitoring elements can be installed layer by layer along the dam height and thickness on-site to collect the central temperature, surface temperature, and internal stress in real-time. When the monitoring data approaches the critical indicators, the flow rate of cold water can be increased simultaneously, the insulation of the concrete placement surface can be strengthened, and the interlayer interval can be extended.

4 Conclusion

This paper takes RCC in water conservancy projects as the subject and conducts model tests on full-process temperature control optimization and crack prevention. Traditional condition (G0), mix proportion optimization condition (G1), mixing temperature control condition (G2), pouring technique condition (G3), internal cooling condition (G4), post-curing condition (G5), and full-process integrated condition (G6) were set up. The temperature control effectiveenss, crack resistance performance, and stress levels of the specimens under the different conditions were compared. It was found that a single temperature control measure could improve the temperature control effect and crack resistance of RCC, but the amplitudes and paths of action were different. Among them, G2 had a prominent effect in reducing the internal temperature peak, and G3 performed well in reducing the temperature difference and delaying cracking. G4 achieved a fast cooling rate but was prone to secondary temperature difference, and its crack resistance effect was limited when used alone. G5 showed the least influence on both temperature control and crack prevention. The full-process integrated optimization achieved low temperature, small temperature difference, and slow cooling, thereby minimizing the risk of cracking to the greatest possible extent.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Federica Buccino, Fondazione Politecnico di Milano, Italy

Reviewed by: MengNan Shi, Sichuan University, China

Liye Zhang, Research Institute of Highway, China