Systems Approach to Concrete Technology
Concrete and the structures constructed from concrete have historically been viewed primarily as the linear sum of their individual parts. Analyses of the interrelationships of the constituent materials of concrete were typically empirical and did not consider the complex chemical and physical interactions between the constituents. Similarly, a systems approach has not been rigorously used to investigate the interactions between concrete and the other parts of engineered structures. The committee has concluded that major advances in the development and application of nonconventional concrete technologies require the use of a MSE systems approach. This chapter examines the MSE systems approach and discusses the need for a model-based design methodology within the construct of total life-cycle costs (i.e., not simply from the perspective of the lowest cost for original construction).
The MSE systems approach can be used to enhance the performance of concrete in two ways. First, conventional concrete can be controlled and manipulated during processing to produce optimal properties. Second, concrete itself can be “compositionally” redesigned to make the concrete less sensitive to process parameters that are difficult to control while yielding superior performance.
In a systems approach, concrete must be viewed as a totally interactive and complex system. This system must be considered over all length-scales, meaning that one must be concerned about the
constituents of the cement, the constituents of the concrete, and the assembly methods used to create the entire engineered structure. This implies that one needs to be concerned with materials and their behavior, and the models that describe them, over perhaps 10 orders of magnitude of length, from less than a nanometer to more than a meter (Figure 1–4). One significant challenge will be to extend the models so that each addresses a different length scale. Another major challenge in modern condensed matter physics, and one that has yet to be completely met even for simple metallic, ceramic, and biological systems, is the melding of these various models in order to understand the effects of the different length scales on the entire engineered structure. The systems approach also needs to consider the entire system as it is manufactured as well as considering its performance after it is put into service. For example, interactions between the concrete and the gravel subsurface or steel structure, as well as interactions between the completed structure and vehicular traffic, must be considered in a systems approach. Finally, the concrete system must be designed to avoid or eliminate the durability problems experienced by conventional concrete and those potential degradation mechanisms introduced by the inclusion of new materials.
At the atomic length scale, the role of the electronic structure of the elements comprising the cement is important. The electronic structure controls the ionic and molecular behavior of cement and therefore is central to the gelation and hydration process. The possibility of molecular engineering of cement by the appropriate and intelligent introduction of other basic elements is an option that should not be cavalierly dismissed. The elements and their amounts that must be added to cement to modify its behavior at any point in time cannot be determined experimentally because of the large number of possibilities. However, molecular dynamics calculations—although probably unable to predict the exact changes in performance that would occur due to the addition of particular elements—could be used to guide experimental design work and make the problem tractable in the laboratory.
On the nanometer length scale, the inherent lower limit of the size of atoms means that structure and processes would have to be considered differently from larger length scales, in which matter can be regarded as continuous.
On the molecular level, particles grow by various random processes that generate fractal structures. There is then a crossover to different growth processes involving the impingement of the fractal cluster, which in turn gives different fractal structures.
At the molecular to micrometer length scale, the role of the size of feedstock particles and fine aggregate on rates of interfacial reactions must be considered. One may be able to engineer and tailor the gelation reaction, for example, by manipulating the size and morphology of the cement particles before water is added or perhaps even by using a controlled distribution of particle morphology so that one part of the structure has different gelation kinetics from those of another. Historically, in the structural hierarchy, this regime has been one of the most difficult to study, and it presents a significant and potentially rewarding research opportunity.
The systems perspective must also be extended to larger length scales. Changing the aggregate, for example, produces a first-order effect related to the properties of the aggregate itself, but the system is also influenced by the nature of the interface between the aggregate and the cement matrix. As another example, silica fume is increasingly being added to make a higher-strength or less permeable concrete. This can also reduce the material's toughness, however, resulting in a material that is more brittle and requires more care during emplacement to prevent cracking. In a systems approach, all of the materials and unit processes used to create a structure would be considered synergistically and not independently.
Finally, the system perspective must be extended to the engineered structure itself. This implies that one needs to consider interactions between the gravel roadbed, the steel superstructure (if any), the concrete, and minor structural components (e.g., expansion joints).
The systems approach must also be implemented over all length scales within two different contexts. First, it must be implemented during the fabrication of concrete and the construction of the structure. The systems approach will optimize jointly the manufacturability of the system and its cost. Optimal combination of manufacturability and cost does not necessarily mean minimizing either one. One special advantage of concrete relative to other manufactured materials is the longer length of time required for processing (i.e., initial mixing through transport to pouring), which offers ample opportunity to adjust material composition and mixing conditions. Second, the systems approach should optimize the performance and reliability of the concrete and the structure. The goal of the optimized system is to minimize the life-cycle costs of the structure by creating an inherently longer lifetime and minimizing the need for maintenance and repair.
The benefits of the systems approach cannot be realized haphazardly. The methods needed to achieve an optimized system may
not be intuitively obvious, as human intuition may be unable to quantitatively comprehend the interactions and influences among all of the materials in these structures over 10 orders of magnitude of length. As a consequence, the key to realizing a systems approach to the materials, the processes, and the structures created from concrete lies in the ability to model the different facets of the entire system rigorously and to meld these models in order to achieve a reasonable understanding of the studied phenomena in the broader context.
Model-based design is a new approach now being used in the manufacturing industry. By definition, it is simply the designing of new materials, structures, and systems quantitatively rather than intuitively. The quantitative nature of the concept allows all pertinent information to be considered simultaneously and synergistically. Each aspect of the design process (i.e., materials and process knowledge) must be rigorously and quantitatively understood, as must the first and higher order interactions between each component of the process. In other words, model-based design uses the right quantitative information at the right place and at the right time. Done properly, model-based design will produce higher-quality products faster and cheaper, especially when considered within the context of life-cycle cost. The concept of model-based design within the context of a systems approach is not a new concept. It has been slowly evolving as computational power increases and is being increasingly applied to the fabrication of “high-tech” systems, including weapons systems, aircraft, and automobiles (NRC, 1988, 1989, 1993, 1995a, 1995c).
The model-based design concept is the next logical step in the systems approach. In order to apply the concept, the materials and the processes used to fabricate the final product must be understood in detail. This complex and quantitative understanding encompasses the fabrication of the material and the structure as well as the performance and long-term reliability of the material and structure. Ultimately, it must be understood that these two aspects of model-based design —the synthesis of the materials and the fabrication of the structure, and its performance and reliability—are themselves coupled.
The first consideration in the use of model-based design is the design of the structure and the processes used to construct it. In such complex cases as a structure fabricated from concrete, sufficient detail
may range from the subtle physics and chemistry of interfacial interactions at the atomic level to the macroscopic behavior of the elastic deformation of the steel tendons used in prestressing. Figure 5-1 shows the interplay of the various aspects of model-based design. Fundamental understanding of the materials and processes will allow the processes to be optimized, but it is necessary to model the process mechanistically and perform computational simulations to reduce the number of experimental attempts that must be made to optimize the process. Experiments should be used to validate the process model and not to optimize the process directly. Once an optimal process has been designed, it is possible to apply intelligent control to the process. To maintain control of the process, it is necessary to monitor explicitly the changes occurring to the materials (in the case of concrete, the temperature, viscosity, pH, etc.) and its microstructure, which requires a system of appropriate sensors. Overlaying the intrinsic behavior of the material during processing is the influence of the external environment. For example, the experimentation required to determine strength development, rates of ingress of chlorides, and resistance to freezing and thawing for each concrete mix design takes several months, and these tests must be repeated as the mix is modified. Model-based design would allow the effects of each mix parameter to be evaluated quantitatively prior to experimentation.
The second issue in model-based design is predicting the performance and reliability of the final structure. This requires a quantitative understanding of the performance of the materials and the structure, based on fundamentals to the maximum extent possible and minimizing reliance on observations of empirical behavior. The impact of the environment on performance and reliability is perhaps even greater than its impact on the original processing due to the much longer time periods involved. Coupling the role of the environment with a quantitative understanding of reliability (e.g., explicit description of how salts degrade reinforced concrete) can be used to develop a fundamental understanding of degradation and to construct appropriate models of the entire system. Empirical knowledge is generally insufficient because it is not possible to perform empirical experiments on the full system under every condition. But understanding, for example, how chlorides penetrate the concrete and attack the reinforcement fundamentally allows one to combine higher-order effects computationally in model-based performance and reliability assessments, including requirements for earthquake zones, fire resistance, lightweight applications, etc.
The performance criteria for nonconventional concrete must be based in part on its interaction with companion materials. For example, earthquake loading is resisted by design of steel reinforcements and related connections that are compatible with the properties of the concrete. These interactions would have to be optimized to determine the required performance of nonconventional concrete.
The ultimate contribution of model-based design within a systems approach is to allow the interactions within and among all of the constituents over all length scales to be quantitatively modeled, permitting an understanding of a particular phenomenon in the broader context. It would be impossible to explore experimentally all possible variables of the materials and processes used to create the structure and then to measure all possible performance variables. Given adequate models, however, it is possible to explore computationally the effects of most, if not all, parameter interactions and to identify specific validation experiments.
The replacement of existing systems with new materials must take into account the conventions that have developed over time to normalize secondary design factors. When new systems are developed, these factors must be identified and addressed.
To realize a model-based approach to concrete and the structures created from it, several key enablers are required. First and foremost, a more complete understanding of all structural features and chemistries is needed at all length scales of the materials and processes used to create the concrete and the engineered structure. This implies that far more extensive research into the theoretical interrelationship among synthesis and processing, structure and composition, properties, and performance of concrete materials is required (Figure 1–1). Second, given more extensive materials and process knowledge, better mechanistic computational models are needed to understand the behavior of the materials and structures during fabrication and in-service. Third, the concept of system optimization through intelligent processing or intelligent systems requires a better suite of capable and inexpensive sensors and actuators.
KEY ENABLERS FOR THE APPLICATION OF A SYSTEMS APPROACH
concrete-base structures design is more extensive fundamental knowledge of materials and processes. As discussed in earlier chapters, the fundamental nature of the physics and chemistry of the gelation process is yet to be understood, the fundamental interfacial interactions between cement and aggregate have not been determined, and the physics and chemistry of the interactions between cement and reinforcement have not been explored in quantitative detail. The constitutive properties (e.g., mechanical behavior) of the cement, aggregate, reinforcements, and their interfaces have not been determined either. Without this fundamental understanding, any significant advance in the science of concrete and its applications will not be forthcoming. Model-based design within the construct of a systems approach will not happen until the fundamental knowledge base is significantly expanded.
Fundamental Issues and Computational Models
If one assumes that the required fundamental materials information described above is available, the next issue is how to use this information to optimize the fabrication and manufacturing processes and to enhance the properties and performance of the final structure. Once again, optimization requires the development of mechanistic models and computational tools that span all length scales, from atomic/molecular to macroscopic. At the most fundamental level, understanding of the electronic/atomic/molecular level is needed. Although it might seem abstract to investigate materials behavior at the atomic and molecular levels for concrete, such information is the key to modeling the interfaces and related interfacial behavior within gels and between gels and aggregate. The models are required for length scales far smaller than a micrometer.
Microstructural evolution models are also required to understand the more macroscopic nature of hydration and gelation. This will be the key to understanding the formation of the various crystalline phases that form in concrete as it sets. In the absence of closed-form physical models of microstructure evolution, computational materials-science methods have been applied (Bentz et al., 1994). This involves treating each particle of material in the microstructure as a cellular automaton. The growth or transformation of each particle during a time cycle is determined by a set of simple transition rules that are based on chemical reactions.
Finally, continuum models are needed for finite-element models to understand the macroscopic aspects of flowing unset concrete, the heat transport in setting concrete, and the elastic/plastic/fracture behavior of the set material and the engineered structure. Although such models have been infrequently applied to concrete, they have been widely used in the study of metals, ceramics, polymers, semiconductors, and the structures created from these materials, as well as composites from various groups of these materials. Once the fundamental materials knowledge is available, modeling will follow naturally. Given that concrete systems and the structures created from them are more complex than the materials and structures for which these models have been successfully used, increased computational power will probably be needed. However, advances in computational engines are occurring rapidly. Continuum models have been run under massively parallel computational architectures with ever-increasing speeds. Terraflop 1 computing will be achieved in the very near future, with three and ten terraflops soon thereafter; and pentaflop computing may well be less than a decade away. Given this massive computational ability, it is only a matter of time before all pertinent models can and will be run.
Many of the parameters used in these models are not discrete but are described by probability distributions. As such, the problem becomes stochastic in nature and must be solved repeatedly to produce a statistically sound solution. Such problems are also very amenable to solution through parallel computation.
Smart Processing and Intelligent Systems
As stated in chapter 3 and chapter 4, sensors will be required to fulfill the vision of intelligent or smart processing as well as the vision of an intelligent system, which in this case would be an intelligent highway or bridge (Davis et al., 1996). Rapid advances in materials knowledge and microelectronics technology have revolutionized the world of sensors. More and more capable sensors are being built at ever-decreasing size and cost. Modern integrated sensors can be produced in a modern semiconductor device fabrication facility at a per-device incremental cost of as low as a few cents. Low cost produces the
1012 arithmetic operations per second.
option of using sensors to assess the processing (and allow feedback control) that could simply be left in the concrete after their work is done. A multitude of properties can be readily measured with embedded sensors, including temperature, viscosity, pH, and organic and chemical contaminants. Devices can be built that measure a number of these parameters, digitize and preprocess the data using on-chip processors, and then communicate the information to the outside world through a network addressable architecture. Several examples of modern sensors that would be particularly sensitive to parameters that might change in concrete during processing and setting are discussed in Appendix B.
Much of what has been discussed above with respect to sensors has dealt with sensors to monitor the concrete as it is being mixed or cured. Sensors might also be used to provide data on roads and bridges while in use. Such sensors would permit damage to be observed at its earliest stages and repairs to be made at minimal cost. Once again, advanced computational models would be needed. The notion of self-aware systems is also not novel, having been explored for weapons systems, aircraft, and, albeit in a more simplistic way, automobiles.
The same technology that is used to fabricate microsensors and microelectronics could be used to fabricate microelectromechanical systems (MEMS). MEMS-based actuators could then be used to control the injection of chemicals into the structure to control or mitigate corrosion and other degradation.
The vast array of sensors now becoming available can produce large quantities of data. Intelligent use of this data requires advanced processing techniques that can include neural networks, expert systems, etc.
To implement a model-based design methodology, concrete and concrete structures must be viewed within a total life-cycle cost context. Although the use of sensors and on-line computational tools to adjust materials, processes, and in-service performance would be expensive, this approach will greatly increase the life of the material and structure and thereby potentially reduce total life-cycle costs. For this approach to succeed, the costs of structure must not be viewed just from the perspective of the lowest cost of original construction.
A favorable trend currently evident in major engineering and construction projects is the increased use of net-present-value estimates over the lifetime of a project in order to provide realistic justification of higher up-front costs and quantify the downstream savings provided by innovations in materials or technologies.
This chapter has attempted to clarify the need for concrete to be considered within the context of model-based design. The methodology is one predicated on a solid understanding of the materials at all length scales and their behaviors from the processing of the initial raw materials to the preparation of actual concrete to be used for pouring or casting, to the setting of the material, to the understanding of the behavior and durability of the concrete during its in-service life. A materials understanding at all length scales means that the behavior of the material is understood quantitatively from the atomistic level to the microscopic level, and on up to the macroscopic or continuum level. This understanding allows one to control the material and its processing continuously.
The implementation of model-based design relies extensively on a sensor-rich environment to allow the development of an explicit quantitative understanding of the materials, the processes, and the degradation mechanisms. Monitoring of the materials and processes would allow for computer-controlled adjustments during placement. Monitoring of the final product in-service would help determine degradation mechanisms and permit the accumulation of data that could be fed back into the model-based design methodology as well as allow cost-effective, localized repairs that would minimize maintenance costs while maximizing in-service time by eliminating routine preventative maintenance. Such a system is often called “smart,” “intelligent,” or “self-aware.”
The technology discussed above, when considered incrementally, is not inexpensive. The materials data that needs to be generated, the models that need to be created, and the sensors that need to be developed will require substantial investment. The use of these enabling technologies to design and construct roadways and bridges and to monitor them during their life will also increase costs. However, there is every likelihood that building the structure using smart processing embedded in a model-based design methodology and
building the structure so that it too is “smart,” “intelligent,” or “self-aware” should significantly extend the life of the structure, greatly minimize repair costs, reduce the amount of time the structure is taken out of service for repairs during its life, and thereby reduce lost opportunity costs. This is a large and complex issue, since the “systems approach” paradigm requires that the discussed technology enablers be considered over all length scales for the cement, the concrete, and then the structure itself. Hence, while the initial investment to build the structure may be substantially greater, the annualized cost of ownership should be significantly reduced. The costs of designing, constructing, monitoring, and maintaining the structure must be summed and amortized over the design life. This yields the predicted annualized life-cycle cost and should be the key financial information given in the original bid.