The current CIKS integrates a number of previously developed and new computer models into a single coherent system. The main menu provided upon accessing the system is shown in Fig. 1. A logical starting point for considering the service life of a concrete structure is the mixture proportioning process. With this in mind, the current ACI guidelines for proportioning ordinary strength (ACI 211.1-91 [4]) and high-strength (ACI 211.4R-93 [5]) concrete have been computerized using a combination of HyperText Markup Language (HTML) forms and CGI programs written in the C programming language. Upon selecting menu item 1, the system user is presented with the forms shown in Figs. 2 through 4 for trial proportioning a normal concrete mixture and specifies the needed parameters and data according to the appropriate ACI guidelines [4,5].
Figure 2: Part 1 of an HTML input form for the mixture propotioning of a conventional concrete.
Fig. 3. Example results of submitting the form in Fig. 2, showing trial mixture proportions and predicted chloride ion diffusivity and thermal properties.
Figure 4: Part 3 of an HTML input form for the mixture propotioning of a conventional concrete.
The choices for the boxes with a button shown in Figs. 2 to 4, based on ACI 211.1-91 [4], are as follows:
The form for a high-strength mixture [5], menu item 2, is similar to that shown in Figs. 2-4, with the following exceptions: aggregate surface property, construction type, air entrainment, and exposure condition are not included; slump must be specified; the use/absence of a high-range water-reducing agent is specified; and the target strength is specified either after 28 or 56 days of curing. Once all parameters have been specified, the user simply clicks on the "Submit form to determine mixture proportions'' button and the resultant trial mixture proportions are returned, as illustrated in Fig. 5, following execution of the mixture proportioning program, written in the C programming language. Based on the trial mixture proportions, the program also returns a predicted value for the chloride ion diffusivity (D) of the in-place concrete [6] and its maximum expected temperature increase under adiabatic (no heat loss) conditions [7].
Figure 5: Illustration of the two-state cyclic "square-wave" function used in characterizing an exposure environment.
The prediction of chloride ion diffusivity from mixture proportions is based on a statistically designed computer experiment which identified water-to-cement (w/c) ratio, volume fraction of aggregates, and degree of hydration as the three major variables influencing the diffusivity of a conventional concrete mixture without mineral admixtures [6]. From the results of this computer experiment, an equation was developed for estimating chloride ion diffusivity coefficients using these three variables. In addition, the CIKS also returns an estimate of the 90% confidence limits for the estimated D value, based on regression of the developed equation to the computer experiment data. Values for w/c ratio and volume fraction of aggregates are directly available from the trial mixture proportioning process. The long term degree of hydration is estimated as 90% of the theoretical maximum achievable hydration, based on the w/c ratio. For w/c ratios greater than or equal to 0.42, there is sufficient capillary porosity for all of the cement to react so that this theoretical maximum is 1, while for lower w/c ratios, this theoretical maximum degree of hydration is given by [(w/c)/0.42] [8]. Alternatively, within the CIKS system, a separate form (menu item 3) exists to estimate the chloride ion diffusion coefficient along with its 90% confidence limits, given user inputs for the w/c ratio, volume fraction of aggregates, and expected degree of hydration, as shown in Fig. 6.
Figure 6: HTML input form for predicting chloride ion diffusivity based on mixture proportioning and anticipated hydration.
Once a value of D has been estimated, it can be employed in a model to predict the service life of a reinforced concrete structure exposed to an external source of chlorides. The simplest approach to this problem, implemented as menu item 4, is to use Fick's second law and solve for t in the following equation [9]:
where Ccorr is the concentration of chloride ions needed at the reinforcement to initiate corrosion, Cext is their external concentration, x is the depth of the reinforcement, D is the chloride ion diffusivity, t is the predicted service life, and erfc(x) = 1 - erf(x) .11] have modified this approach slightly to consider the statistical variation in the depth of the reinforcement bar (assuming a normal distribution characterized by a mean and standard deviation), employing chloride ion concentrations reported in mass of chloride per unit volume of concrete, and taking Cext to be the chloride ion concentration measured at a depth of 12.7 mm (0.5 in.). The approach outlined in their report has been implemented in the current CIKS. An alternative to the simple erf solution of Fick's second law is to employ a one-dimensional finite difference solution [12], which directly incorporates the time-dependent variability of the exposure environment and the performance differences between the bulk and surface layer concrete. The model, menu item 5, developed in the present research allows for the following parameters to be specified by the user:For the case where the concrete diffusivity can be described by a two-layer model and all others of the above effects can be ignored, an analytical solution for the concentration profile as a function of time has been obtained by Andrade et al. [15], based on the solution originally derived by Carslaw and Jaeger in terms of heat transfer variables [16]. This solution can also be viewed within the CIKS by selecting main menu item 6, Advice on analyzing chloride ion penetration profile data. The current model in the CIKS considers only diffusion under saturated conditions; it should be noted that comprehensive models for diffusion into partially saturated concrete have been previously developed by Saetta et al. [17]. Although much more computationally intensive, models such as that of Saetta et al. may also someday be executable over the WWW.
Once a user specifies all of the above parameters and submits the form for this module, the underlying C program returns a plot showing the predicted total and free chloride ion concentrations as a function of depth. Knowing the depth of the reinforcement and the chloride concentration necessary to induce corrosion, the user can then determine if the chloride ion concentration at the reinforcement is such that corrosion will be probable after the specified exposure time. In addition to returning this plot, the program also optionally sends a file, containing a numerical listing of the inputs and results, by e-mail to a user-specified address.
