Head of Team : Dr.rer.nat. Ir. Lilik Eko Widodo, MS.

Team Members : Dr. Ir. Sudarto Notosiswoyo, M.Eng. ; Irwan Iskandar, ST., MT., Ph.D. ; Zuher Syihab, ST., MT., Ph.D. ; Tedy Agung Cahyadi, ST., MT.

Introduction

Mine drainage is a part of mining activities. This activity becomes a priority, however, if the mining activities encounter problems in managing water, particularly groundwater. The presence of groundwater in an open mining front may cause problems such as slope instability due to rising groundwater levels (Kong et al. 2008). To overcome groundwater problems for slope stability, a mine drainage system is needed, for example installation of drainholes on the slope. A drainhole is a draining method which installs vertical, horizontal or inclined holes on a slope, which discharge or drain groundwater from the slope by gravity drainage. Drainhole installation is a complex problem if it is applied in fractured groundwater flow media (Cahyadi et al. 2016). Current installations of drainholes in an open pit mine with fractured groundwater flow media still use a uniform pattern (Silaen et al. 2011; Leech and McGann 2007). Rahardjo et al. (2003) stated that a low number of drainholes placed in the correct locations based on a conceptual model would be more effective than a higher number of drainholes with uniform spacing. The uniform spacing of drainholes may cause some drainholes not to function, which will also raise the cost of installation and operation (Cahyadi et al. 2016). This research attempts to provide a new approach to mine drainage systems by setting drainhole parameters, or decision variables, such as the number, location and length of drainholes, leading to optimal groundwater drainage from an open mine slope, utilizing the Multistage VmodGA SO from a conceptual model perspective. The conceptual model will be further developed as a framework for managing groundwater flow in fractured GW flow media in mine slopes.

Methods

In this research, the number or amount, the location or position, and the length of the drainholes represent the chromosomes. The purpose of SO is to maximize the decrease or the drawdown of the groundwater level by minimizing the number and length of the drainholes. To prove whether the Multistage VmodGA SO succeeds, cross-validation is performed. GA optimization is a stochastic optimization method which copies evolutionary and natural selection principles to attain an optimal value (Rao 1978). Evolution is a natural selection process based upon the quality of living creatures; superior creatures have a better chance of surviving and passing their qualities on to the next generation. According to similar principles, the GA searches for an optimum solution by attempting “evolution” towards possible solutions. Only a superior solution with a high fitness value has a chance to survive within a solution population. GA states one solution in the form of chromosomes, which has 8 binary (0 and 1) codes.

Results

The first step of the multistage VModGA SO is to look for the minimum number of drainholes and the corresponding optimum locations. The second step is to find the minimum drainhole length. In the first step, the objective function is limited by the groundwater drawdown, the drainhole spacing and the number of drainholes permitted within a conceptual model domain of 150 m x 150 m x 150 m.

The first step is performed within several generations until the optimum drainhole locations are found. Drainhole candidates, i.e. the chromosomes which always appear in each generation, represent the optimum drainhole candidates in the corresponding generation. The Multistage VModGA SO process with five generations takes about eight hours, resulting in optimum drainhole candidates at the coordinates (40, 126, 133), (40, 75, 133) and (40, 26, 133), which lead, respectively, to maximum groundwater drawdowns at piezometer No. 2, No. 5 and No. 4, as explained above. Chromosomes in the fifth generation result in 10 possible optimum drainhole locations, with limited spacing between drainholes greater than 40 m. Accordingly, the three optimum drainholes inducing the most groundwater drawdown were chosen.

The second step is to determine the minimum length of an individual drainhole. If the first step is associated with chromosomes containing genetic information for the locations or the coordinates of drainhole candidates and the related groundwater drawdown, then the second step is associated with chromosomes containing genetic information for the length of individual drainhole candidates. In the second step, the GA process is performed following the process of selection, crossover, mutation and elitism, following the same principles as in the first step. In the first step, chromosomes (X, Yn, Z) are simulated using Visual ModFlow to result in optimum drawdowns (∆Hn, X, Yn, Z). On the other hand, in the second step, chromosomes (40, 126, 133, L1), (40, 75, 133, L2) and (40, 26, 133, L3) are simulated using Visual ModFlow to produce groundwater drawdowns at three drainhole candidates as follows: (∆H1, 40, 126, 133, L1), (∆H2, 40, 75, 133, L2) and (∆H3, 40, 26, 133, L3).

Finally, GA produces the following optimum drainhole candidates: (∆H1=8.94 m, 40, 126, 133, L1=97m) for drainhole 1, (∆H2 = 7.15 m, 40, 75, 133, L2 = 80 m) for drainhole 2 and (∆H3 = 6.70 m, 40, 26, 133, L3 = 87 m) for drainhole 3. To attain convergent drainhole lengths and groundwater drawdown values, a sum of the total length and total drawdown must be computed. The total length of the drainholes starting from the initial generation to the fourth generation results in a value which tends to rise until it reaches a similar value. This is similar to the process producing the total groundwater drawdowns from three observation wells. This condition proves that the GA process has gained a convergence value. The convergence of the optimum lengths of three drainholes can be seen in Figure 5. The second step takes 11.25 hours; thus, the total time of the multistage VModGA SO multistage process from step 1 to step 2 takes 19.5 hours.

Conclusion

1. In this research, the Multistage VModGA SO was developed to determine the optimum drainhole candidates in terms of number, location and length from a conceptual model perspective.
2. The Multistage VModGA SO solution process is very efficient, as it requires a shorter runtime than other methods.
3. With the optimum drainholes, the installation of drainholes with uniform spacing should not be performed in the field, because it is inefficient, ineffective and uneconomical.
4. The drainhole optimization using multistage VModGA SO is very sensitive and depends on the accuracy of the hydraulic parameter model of the groundwater flow media, especially fractured groundwater flow media, which is heterogeneous and anisotropic. Therefore, for the Multistage VModGA SO method to be accurate, the hydraulic parameter distribution model of the groundwater flow medium first needs to be verified and validated.