Comparative study of mine dewatering control systems

Stéphan Taljaard, Johann F. van Rensburg, and Hendrik J. Groenewald

 sufficient training and knowledge of the pumping system,

Abstract—Deep-level mines use a series of dams and pumps as well as its requirements and constraints. Dewatering
known as a dewatering system to remove water from
systems can also be controlled automatically by
underground. Different methods exist for automatically
controlling dewatering systems. The three predominant methods computerised systems (known as “automated control”).
used in deep-level South African mines are programmable logic Automating the control of the dewatering system has
controllers (PLCs), supervisory control and data acquisition advantages, such as improved safety of the people and
(SCADA) and third-party control software. However, a literature equipment in and around the pumps [1], cost savings
survey indicated a lack of comparative studies on the three through demand-side management (DSM) initiatives [1],
predominant control methods. A feature and control performance reduced labour and operating costs [2] and reduced
scoring method was therefore developed to experimentally
compare the performance of different control methods. This maintenance because of lessened pump on/off
control performance score takes into consideration the energy switching [1].
cost for controlling the dewatering system. The performance Typically, automated control of pumping systems is
scoring method was applied on three case studies on South performed by the following control systems:
African mines, with each case study having a unique dewatering • Control via programmable logic controllers (PLCs):
system that differs in complexity from the others. It was found
Pumps and instrumentation are wired to the PLCs. The
that the control of complex dewatering systems could be better
optimised by using control systems that offer the capability to pump control algorithm is programmed into the PLC.
incorporate more complex/site-specific logic into the control • Control via supervisory control and data acquisition
algorithm. Complex dewatering systems require a non-generic (SCADA) systems, with PLC implementation:
approach. For simpler dewatering systems, higher complexity The pump control algorithm is programmed in the
control systems are not strictly required, however they might SCADA. The SCADA issues start/stop commands to
offer the advantage of more advanced features which aid in the
implementation and use of said control system. In all three case the underground PLCs, which in turn start/stop the
studies, third-party control software proved to be the optimal relevant pumps.
choice of control system. This was followed by SCADA control • Third-party control software:
and lastly, PLC control. The third-party control software implementation
externalises the pump control algorithm from the
Index terms—Automatic control, Control systems, Mining
SCADA and PLCs. The third-party control software
industry, Pumps.
connects to the SCADA system via a real-time process
1 INTRODUCTION data connectivity standard, such as the Open Platform
Communications standard (OPC) to read various
Deep-level mines use water underground for cooling
pumping system parameters. The parameters are then
and mining operations. Natural water seeping from rock
processed according to the pump control algorithm.
surfaces (known as fissure water) may also be present
The output of the algorithm is fed back to the SCADA
underground. The used mining water and fissure water
to enable the controlling of the pumps.
accumulates in water holding dams on the mining levels.
In order to prevent flooding of the mine, the accumulated
A study of available literature revealed numerous
water needs to be removed with a pumping system (also
studies investigating the optimisation of pump control
known as a dewatering system).
schedules using mathematical techniques, without
A pumping system consists of a series of reservoirs and
focusing on the actual implementation of these techniques
pump stations located on multiple mining levels. The
[3]–[10]. Furthermore, one finds multiple studies doing
mining and/or fissure water is then pumped between the
optimisation using a single control system only [1], [11]–
mining levels until it reaches the surface.
[26]. However, no literature was found that contains an
Mine dewatering systems can be controlled manually
experimental comparison of different pump control
by pump attendants at the pumping stations and/or by
systems or approaches. A need therefore exists for a
control room operators if the pumps are remotely
thorough evaluation of each of these control systems in
controllable. This however requires the operators to have
order to be able to determine which will be the optimal

This work was sponsored by ETA operations. J. F. van Rensburg, North-West University, CRCED-Pretoria,
S. Taljaard, North-West University, CRCED Pretoria, Pretoria, South Pretoria, South Africa (e-mail: [email protected]).
Africa (e-mail: [email protected]). H. J. Groenewald, North-West University, CRCED-Pretoria, Pretoria,
South Africa (e-mail: [email protected]).

Authorized licensed use limited to: National Institute of Technology Karnataka Surathkal. Downloaded on January 09,2025 at 05:26:36 UTC from IEEE Xplore. Restrictions apply.
 choice for the control of dewatering systems of varying 2.2 Scoring methodology
complexities. A scoring methodology was developed to compare the
The aim of this research is to develop and implement a different control systems. The developed scoring consists
method for the comparison of different mine dewatering of two main scores: a “performance score” and a “feature
pump control systems. More specifically, the objectives are score” that are combined into a final score out of 100.
to:
2.2.1 Scoring methodology: performance score (PS)
i. Compare the three control systems (PLC, SCADA and
third-party control software) through: The performance score describes how well the control
system can control a dewatering system. The score consists
a. Comparison of the performance of the control
of three parts, each considered as important when
systems (by means of simulating the output of the
controlling pumping systems:
control system).
b. Evaluation and comparison of control system i. Cost score (CS)
features that improve ease-of-use and/or are Using the Eskom tariff structure applicable to the site,
beneficial to the control of the dewatering system. the electricity cost of the power profile can be calculated.
ii. Draw conclusions as to how the complexity of the This is called the “simulated profile cost” (“SPC”). Next,
dewatering system affects the choice of control system. the “best profile cost” (“BPC”) is calculated – the lowest
energy cost needed to remove the amount of water from the
2 METHODS mine. The cost score (CS) is then calculated by dividing
2.1 Model and simulation development the BPC by the SPC.
To simulate the control of a dewatering system, a ii. Level score (LS)
mathematical model (mass balance) for each dewatering The level score (LS) is an indication of the capability
level was developed: of the control system to maintain the dam levels within
𝑑𝐿𝑓 their control ranges. The level score is calculated as the
= 𝑉𝑐 (𝑄in − 𝑄out ) (1) number of dams that stayed within their penalisation limits
𝑑𝑡
𝑑𝐿
where 𝑓 represents the change in dam level as a fraction divided by the total number of dams.
𝑑𝑡 iii. Pump cycling score (PCS)
of its total capacity with respect to time (𝑡), 𝑉𝑐 represents
“Pump cycling” refers to the starting and stopping of a
the dam capacity, and 𝑄in and 𝑄out represents the
pump in quick succession. This may cause damage to the
volumetric flow of water into and out of the dam.
pump, motor, pipes or other fittings [28], [29].
Equation (1) is solved iteratively. Therefore, an analytical
The pump cycling score (PCS) is an indication of the
equation akin to Euler’s method is used instead of
capability of the control system to schedule the pumps in
differential equations:
100% such a way so as to not have a negative impact on the life
𝐿new = 𝐿old + Δ𝑡 (𝑄in − 𝑄out ) (2) or maintenance cost/frequency of the pump. If no pump
𝑉𝑐
cycling occurred, a pump cycling score (PCS) of one is
where 𝐿 represents the dam level rewritten as a percentage
awarded. If unnecessary/preventable pump cycling
of the dam’s capacity. occurred, a score of zero is awarded.
The model (2) is solved for each dewatering system
considered. An iterative approach is followed for this, Total combined performance score (PS):
where a new set of values is calculated for each dewatering The three scores calculated above are combined into the
system every simulated second (Δ𝑡 = 1). After total performance score as a weighted average:
calculations, control decisions are made according to the Performance 𝑤𝐶𝑆 × 𝐶𝑆 + 𝑤𝐿𝑆 × 𝐿𝑆 + 𝑤𝑃𝐶𝑆 × 𝑃𝐶𝑆
= (3)
control systems’ pump scheduling algorithms (described in score (PS) 𝑤𝐶𝑆 + 𝑤𝐿𝑆 + 𝑤𝑃𝐶𝑆
detail in [27]), ready to be effective the next second. where 𝑤𝑖 represents user-definable weights.
Initial values of the mathematical model are chosen at Due to operational concerns, dams reaching critical
the start of the simulation. A single day is then simulated levels or control systems causing unnecessary wear on
for each case study. The initial values of the mathematical pumps (in the form of unpreventable pump cycling) should
model are the same as the actual values that the respective be heavily penalised. These factors are more important
days start with (dam levels and number of pumps running). than whether or not the control system is able to control the
After verifying whether the assumptions create an accurate pumps using a minimal energy cost. For this reason, the
and representative model of the control system for the weights are chosen as 𝑤𝐶𝑆 = 1, 𝑤𝐿𝑆 = 2, and 𝑤𝑃𝐶𝑆 = 2.
specific day (the “base case” simulation), the influence of This means that the level score and pump cycling score
the control system’s automated control is investigated by carry more weight, thus leading to more pronounced
means of simulation. effects on the total performance score.
The developed model was simulated using the 2.2.2 Scoring methodology: feature score (FS)
programming language Python, along with additional The feature score serves as a means of evaluating each
modules NumPy and Pandas. All source code is available control system’s features and ease of use, as this also plays
electronically on GitHub (http://bit.ly/2rqp9IE). a part in deciding which control system to use and how it
will be used. The following control system features have

Authorized licensed use limited to: National Institute of Technology Karnataka Surathkal. Downloaded on January 09,2025 at 05:26:36 UTC from IEEE Xplore. Restrictions apply.
 been identified as being beneficial when controlling 2.2.3 Scoring methodology: combined total score
dewatering systems. These features were gathered from The total score is a weighted average of the
personal industry experience and from the work done by performance score and feature score and is the eventual
other researchers [2], [30, pp. 364–365], [31, pp. 286–300]. metric with which control systems are compared:
i. Simulation/testing environment: the built-in capability 𝑤𝑃𝑆 × 𝑃𝑆 + 𝑤𝐹𝑆 × 𝐹𝑆
of the control system to test the control of the water Total score = 100 (4)
𝑤𝑃𝑆 + 𝑤𝐹𝑆
pumping system. This enables testing of the control The weights are chosen as 𝑤𝑃𝑆 = 6 and 𝑤𝐹𝑆 = 3, since
philosophy before deployment of control parameters. whether the control system actually works is deemed more
ii. Optimised pumping schedule: the capability of the important than the additional features it offers. It is thus
control system to calculate and/or perform optimised preferred that the performance score carry more weight.
control of pumps based on an optimised pumping
schedule. This schedule should lead to reduced running 3 RESULTS
costs (electricity and maintenance cost) for the system. The developed comparison methodology was
iii. Control and automated operation: the capability of the implemented on three case studies provided by a South
control system to control the components of the African gold mining group.
pumping system in one way or another. Control tasks It may be possible to develop a control system to
should be completed without needing 24-hour human perform similarly or on the same level as another. For
assistance. Emergencies should not cause failure of the example, a SCADA can follow the same control algorithm
control system. as a third-party control software does. However, the
iv. Monitoring and reporting: the capability of the control development could be impractical; whether the actual
system to provide monitoring capability for operators system capabilities, development time, skills required, etc.,
to be able to observe and/or control the process. allows for that, will determine the practicality and
Furthermore, this feature refers to the capability of feasibility of developing the system to such an extent.
automatic logging, managing and reporting of data Because of this, the control systems will be considered as
relevant to the system being controlled. being 𝑥-factored, indicating the number of main control
v. Alarm handling. the capability of the control system to variables being considered by the system. In this research,
raise warnings if process parameters reach certain the PLC implementation is an example of 1-factor control
limits. These alarms attract the operator’s attention to (where only the upstream dam level is considered). The
elicit rapid response to keep the process within control.
example for 2-factor control is SCADA control (which
vi. Skill level required. the skill level required by the considers both the levels of the upstream, as well as
person doing setup or making maintenance changes on downstream dam). Third-party control software is the
the control system. To set up the control system, specific example for 𝑛-factor control in this study.
programming of some sort is required.
1-factor and 2-factor control uses an almost identical
In Table I, each feature is given a score between zero and algorithm, since it is the algorithm used by the case study
three, indicating how well that feature has performed in a mines’ SCADA (Wonderware System Platform
given control system. (ArchestrA)) for pump scheduling. The 1-factor control
system considers only the dam level for the dewatering
Table I. Control system feature score breakdown dams that it is employed for, whereas the 2-factor model
Max. possible also considers the level of the downstream dam(s).
score An industrial pump dewatering simulation and control
Simulation / testing environment 3 software package used in many South African mines, Real-
- Yes = 3, No = 0. Time Energy Management System (REMS), was selected
Optimised pumping schedule 3 for the simulation of 𝑛-factor control.
- No = 0. The authors describe the control algorithms in full
- Follow/recommend, leading to reduced detail elsewhere [27].
electricity cost. If not, -1.
- Workload balancing. If not, -1. 3.1 Feature scoring results
Control and automated operation 3 Feature scoring results are independent of the case
- Able to control = 1 studies (i.e. not site-specific), therefore this scoring will not
- Human assistance not needed = 1
be repeated for each case study. Table II presents the
- Emergency situation handling = 1.
Monitoring and reporting 3
summarised feature scoring results for the different control
- Monitoring = 1. systems. Refer to full control system details [27] for a full
- Logging = 1. explanation as to why each score was awarded.
- Reporting = 1.
Alarm handling 3
- Scoring relative to other control systems.
Skill level required 3
- Scoring relative to other control systems.
18

Authorized licensed use limited to: National Institute of Technology Karnataka Surathkal. Downloaded on January 09,2025 at 05:26:36 UTC from IEEE Xplore. Restrictions apply.
 Table II. Awarded control system feature scores. 30 579.18
1× +2×1+2×1
Subtotals are in bold and underlined [27]. 37 748.91
=
x-factored 1+2+2
control system = 0.9620
1 2 3 The performance score (PS) and feature score (FS) is
Simulation / testing environment 0 0 3 combined into a total score using (4), along with the
- Yes = 3, No = 0. predetermined weights:
Optimised pumping schedule 2 2 3 𝑤𝑃𝑆 × 𝑃𝑆 + 𝑤𝐹𝑆 × 𝐹𝑆
- No = 0.
Total score = 100 (4)
𝑤𝑃𝑆 + 𝑤𝐹𝑆
- Follow/recommend, leading to reduced 9
electricity cost. If not, -1. 6 × 0.9620 + 3 ×
= 100 18
- Workload balancing. If not, -1. -1 -1 6+3
Control and automated operation 2 3 3 = 80.80
- Able to control = 1 1 1 1
- Human assistance not needed = 1 0 1 1 3.2.1.3 2-factor control simulation
- Emergency situation handling = 1. 1 1 1 2-factor control resulted in identical behaviour (and
Monitoring and reporting 1 3 3 results) as 1-factor control. This is because the case study
- Monitoring = 1. 1 1 1 only had one pumping level and the only difference
- Logging = 1. 0 1 1
between the 1-factor and 2-factor control systems was the
- Reporting = 1. 0 1 1
logic related to the handling of downstream dams. The
Alarm handling 1 2 3
- Scoring relative to other control systems. simulation thus did not require the inclusion of dams
Skill level required 3 1 3 downstream from the dewatering level.
- Scoring relative to other control systems. 3.2.1.4 n-factor control simulation
9 11 18
𝑛-factor control was found to perform a marginally
3.2 Performance scoring results better electrical load-shift than the other control systems,
3.2.1 Case study 1: mine with one dewatering level while keeping dam levels closer to the control range. The
BPC was R30 730.13 and the SPC was R37 681.08.
Case study 1 has one dewatering level. Water is
pumped, by four pumps, to holding dams on another level Fig. 1. lists the performance, feature and total scores
located closer to surface. From these dams, water is then awarded to Case study 1, a simple dewatering system.
gravity-fed freely to another nearby mine shaft.
3.2.1.1 Validation of base case simulation
The base case simulation followed the actual pump
running schedules and dam level with great accuracy. The
mean deviation of predicted values with respect to the
observed ones (root mean square deviation, RMSD) of
1.25% supported this. The base case simulation was
therefore deemed valid and further inferences could be
made.
3.2.1.2 1-factor control simulation
Simulation of the behaviour of the 1-factor control
system on Case study 1’s dewatering system exhibited no Fig. 1. Case study 1: total scores.
pump cycling. The dam level was controlled within its
control limits. The best profile cost (BPC) for the operation 3.2.2 Case study 2: mine with two dewatering levels
of this simulated day was R30 579.18. The simulated Case study 2 has two pumping levels, each with two
profile cost (SPC) for this day was R37 748.91. The 3 ML dams. Each level has fissure water inflow that varies
difference between the BPC and SPC indicates cost- in flow rate throughout the day. In total, the dewatering
ineffective control, which will penalise the performance system consists of nine pumps, of which a maximum of
score. three can run simultaneously per level. Water pumped to
For Case study 1’s 1-facor control simulation, the surface accumulates in dams which are designed to
calculation of the final scores will be shown below as an overflow, and these dams therefore need not be included in
example. For subsequent simulations and case studies, the simulation.
calculations will not be shown. Calculation of the
performance score is done using (3) along with the 3.2.2.1 Validation of base case simulation
predetermined weights: The base case simulation followed the actual pump
Performance 𝑤𝐶𝑆 × 𝐶𝑆 + 𝑤𝐿𝑆 × 𝐿𝑆 + 𝑤𝑃𝐶𝑆 × 𝑃𝐶𝑆 running schedules and dam level with great accuracy. The
= (3)
score (PS) 𝑤𝐶𝑆 + 𝑤𝐿𝑆 + 𝑤𝑃𝐶𝑆 average RMSD for the dam level comparison of 0.700%

Authorized licensed use limited to: National Institute of Technology Karnataka Surathkal. Downloaded on January 09,2025 at 05:26:36 UTC from IEEE Xplore. Restrictions apply.
 supports this. The base case simulation was thus deemed overflowed. The remaining three dewatering levels’ dams
valid and further inferences could be made. were controlled within their control ranges. The BPC for
the operation of this day was R185 691.93. The SPC for
3.2.2.2 1-factor control simulation
this day was R258 259.71.
The control system was able to control the dam level
between its minimum and maximum allowable values, but 3.2.3.3 2-factor control simulation
some amount of unnecessary pump cycling occurred. The Control-system induced pump cycling occurred at one
BPC for the operation of this simulated day was of the dewatering levels. Further, all dams were controlled
R91 878.95. The SPC for this day was R121 942.09. within their control limits. The BPC for the operation of
this day was R185 691.93, whereas the SPC was
3.2.2.3 2-factor control simulation
R257 700.40.
Results were identical to the 1-factor simulation.
3.2.3.4 𝑛-factor control simulation
3.2.2.4 𝑛-factor control simulation
No pump cycling occurred. However, the surface dam
The 𝑛-factor control system was able to control the dam overflowed as a strategic choice in order to remove water
level between its minimum and maximum allowable from underground and reduce the flooding risk at the more
values, without preventable pump cycling occurring. The sensitive underground levels. The BPC for the operation of
BPC for the operation of this day was R91 171.81, whereas this day was R192 497.32, whereas the SPC was
the SPC was R121 379.67. R270 042.86.
Fig. 2. lists the performance, feature and total scores Fig. 3. lists the performance, feature and total scores
awarded to Case study 2, an intermediate-level-of- awarded to Case study 3, a complex dewatering system.
complexity dewatering system.

Fig. 3. Case study 3: total scores.

Fig. 2. Case study 2: total scores.
4 DISCUSSION
3.2.3 Case study 3: mine with four dewatering levels
The control system scores, as obtained from evaluation
Case study 3 has four pumping levels, each with one
of the three case studies, are combined and repeated in
5 ML dam usable for water storage. 18 pumps work
Fig. 4.
together to pump fissure water out to the surface, where
overflowing of the surface dam should be avoided. Fissure
water flow rates also vary throughout the day.
3.2.3.1 Validation of base case simulation
The base case simulation followed the actual pump
running schedules. For some of the dewatering levels, the
simulated dam level was higher than the actual dam level.
This is due to the erratic inflow of water into the dams,
which is not perfectly captured by the 30-minute resolution
used for calculations. However, with the simulated dam
levels being higher than the actual dam levels, this means
that the assumptions create a model which errs on the Fig. 4. Summary of control system scores.
conservative side. Nevertheless, Case study 3’s dewatering
system was simulated accurately and is deemed valid, with Case study 1 (CS1) has only one pumping level and the
an average RMSD value for dam levels of 2.670%. downstream dam is not of concern. This dewatering system
3.2.3.2 1-factor control simulation is therefore uncomplicated and it was found that the three
Pump cycling occurred on all dewatering levels. One considered control systems performed virtually identically.
dewatering level’s dam as well as the surface dam also Eventually, what sets the control systems apart for a simple

Authorized licensed use limited to: National Institute of Technology Karnataka Surathkal. Downloaded on January 09,2025 at 05:26:36 UTC from IEEE Xplore. Restrictions apply.
 dewatering system such as CS1, are the features of the increased by making use of control systems that do not
control system. 2-factor control (i.e. the mine SCADA allow complex or customisable control algorithms.
control with PLC implementation) proved to offer more in The 1 factor and 2 factor control algorithm is likely to
terms of features compared to 1-factor control (PLC-only cause unnecessary cycling of pumps. This is due to its
implementation). However, 𝑛-factor control (i.e. third- handling of the case when “two too many pumps are
party control software in the form of REMS) offered even running”. Another contributing factor is the behaviour
better yet. Thus, all things considered, 𝑛-factor control when dam levels reach the lower control limit: dam levels
offers a better choice than 2-factor control, and 2-factor will then be controlled between this dam level (%) and 5%
better than 1-factor. For a simple dewatering system such below that. This default value of 5% is likely to cause pump
as CS1, the requirements of the instrumentation/control cycling because of this narrow control range of 5%.
engineer will therefore eventually be the deciding factor in
5 CONCLUSION
the choice of control system to implement. The
implementation and upkeep costs for the control system A comparison of different control systems was done
will also play a more important role (though this falls out with regards to their utility on the dewatering systems of
of scope for the current study). the gold mining sector. These control systems consisted of
Case study 2 (CS2) has two pumping levels of concern. 1-factor, 2-factor and 𝑛-factor control systems.
This dewatering system is more complex than the For dewatering systems of lower complexity, the
dewatering system at CS1. The difference between the choice of the control system to use was found to depend
performance of the different control systems was therefore mainly on the features required, since differences in control
found to be more pronounced. In terms of performance, system performance was negligible. However, as the
1-factor and 2-factor control were identical, because the complexity of the dewatering system increases, so does the
level of the dam downstream from the bottom pumping need for a control system offering more complexity in the
level did not reach near-full values. 𝑛-factor control control algorithm. Therefore, control of complex
however offered superior performance. The 1-factor and dewatering systems can be better optimised using control
2-factor control suffered in their performance scores, systems offering the capability of incorporating more
because of unnecessary pump cycling induced by the complex/site-specific logic into the control algorithm.
control algorithm. 𝑛-factor control therefore proved to be Using a generic approach, as with the case studies’ mine
the best choice. SCADA algorithm, is not necessarily always the optimal
Case study 3 (CS3) has five levels of concern, four of choice.
which has pumps forming part of this complex mine For the mining group providing the case studies,
dewatering system. It was found that that using 2-factor 𝑛-factor control (in the form of REMS) was shown to be
control instead of 1-factor control offered better control of the optimal choice of control system to use. This was
the dewatering system. Using 𝑛-factor control, however, followed by 2-factor control (in the form of mine SCADA)
offered an even more significant advantage. Even though and then 1-factor (mine SCADA algorithm implemented in
the surface dam overflowed during simulation, 𝑛-factor PLCs only).
control performed the best in terms of removing amounts REFERENCES
of water from the mine and returning operation of the [1] P. J. Oberholzer, “Best practices for automation and control of
dewatering system back to normal. Risk of underground mine dewatering systems,” MEng dissertation, NWU,
flooding was also lowered by using 𝑛-factor control. The Potchefstroom, 2014.
[2] J. W. Rautenbach, “Engineering a novel automated pump control
1-factor and 2-factor control again suffered in their system for the mining environment,” PhD thesis, NWU,
performance scores, because of unnecessary pump cycling Potchefstroom, 2007.
induced by the control algorithm. 𝑛-factor control thus [3] B. L. M. Brion and L. W. Mays, “Methodology for optimal
operation of pumping stations in water distribution systems,” J.
proved to be the best choice considering actual Hydraul. Eng., vol. 117, no. 11, pp. 1551–1569, 1991.
performance, features, and a combination of the scores. [4] L. Ormsbee, S. Lingireddy, and D. Chase, “Optimal pump
It can be extrapolated that, for mine dewatering systems scheduling for water distribution systems,” in Multidisciplinary
more complex than Case study 3, the difference in control international conference on scheduling: theory and applications
(MISTA 2009), 2009.
system performance would be even more pronounced. This [5] M. F. K. Pasha and K. Lansey, “Optimal pump scheduling by
will mean that the use of 𝑛-factor control (REMS) will be linear programming,” in World environmental and water
even more essential. More complex dewatering systems resources congress 2009, 2009, no. 2009, pp. 38–38.
[6] J. G. Bene, “Pump schedule optimisation techniques for water
can, for example, consist of more dewatering levels or distribution systems,” Dissertation — PhD, University of Oulu,
contain energy recovery devices, such as three-chamber Finland, 2013.
pump systems. [7] A. N. Hasan, B. Twala, and T. Marwala, “Predicting mine dam
levels and energy consumption using artificial intelligence
The manner in which the dewatering system is methods,” in 2013 IEEE Symposium on computational
controlled also has an impact on the energy cost required intelligence for engineering solutions (CIES), 2013, pp. 171–175.
for that operation. It was shown that the control system [8] X. Zhuan and X. Xia, “Optimal operation scheduling of a
pumping station with multiple pumps,” Appl. Energy, vol. 104,
influences the manner in which the dewatering system is pp. 250–257, Apr. 2013.
controlled, and thus should affect the required energy cost. [9] M. Behandish and Z. Y. Wu, “Concurrent pump scheduling and
With dewatering systems of higher complexity having a storage level optimization using meta-models and evolutionary
algorithms,” Procedia Eng., vol. 70, pp. 103–112, 2014.
larger need for complex control, the energy cost is

Authorized licensed use limited to: National Institute of Technology Karnataka Surathkal. Downloaded on January 09,2025 at 05:26:36 UTC from IEEE Xplore. Restrictions apply.
 [10] V. Puleo, M. Morley, G. Freni, and D. Savić, “Multi-stage linear Johann F. van Rensburg is a
programming optimization for pump scheduling,” Procedia Eng., registered professional engineer and
vol. 70, pp. 1378–1385, 2014. holds a PhD in electrical engineering
[11] S. Pezeshk and O. J. Helweg, “Adaptive search optimization in
from North-West University. He is a
reducing pump operating costs,” J. water Resour. Plan. Manag.,
vol. 122, no. 1, pp. 57–63, 1996. lecturer at the NWU’s Centre for
[12] W. F. Van Niekerk, “The value of simulation models for mine Research and Continued Engineering
DSM projects,” MEng dissertation, NWU, Potchefstroom, 2013. Development (CRCED) in Pretoria.
[13] W. J. J. Breytenbach, “Integration of electricity cost saving
interventions on a water distribution utility,” MEng dissertation,
NWU, Potchefstroom, 2014.
[14] T. Smith, “Automated control of mine dewatering pumps,” MEng Hendrik J. Groenewald is a
dissertation, NWU, Potchefstroom, 2014. registered professional engineer and
[15] C. Cilliers, “Cost savings on mine dewatering pumps by reducing holds a PhD in Electronic and
preparation-and comeback loads,” MEng dissertation, NWU, Computer Engineering from North-
Potchefstroom, 2014.
West University. He is enrolled for
[16] H. L. Grobbelaar, “Maintenance procedures on DSM pumping
projects to improve sustainability,” MEng dissertation, NWU, post-doctoral studies with CRCED-
Potchefstroom, 2014. Pretoria, a division of North-West
[17] J. P. De Jager, “Investigating the effect of pump availability on University.
load shift performance,” MEng dissertation, NWU,
Potchefstroom, 2015.
[18] M. Van der Merwe, “Strategies to revive DSM mine pumping
projects under the new ESCo model,” MEng dissertation, NWU,
Potchefstroom, 2016. Presenting author: The paper will be presented by S. Taljaard.
[19] B. Dieu, “Application of the SCADA system in wastewater
treatment plants,” ISA Trans., vol. 40, no. 3, pp. 267–281, 2001.
[20] G. Cembrano, J. Quevedo, M. Salamero, V. Puig, J. Figueras, and
J. Martí, “Optimal control of urban drainage systems: a case
study,” Control Eng. Pract., vol. 12, no. 1, pp. 1–9, 2004.
[21] R. P. Richter, “Comparison between automated and manual DSM
pumping projects,” no. November, 2008.
[22] K. G. Shankar, “Control of boiler operation using PLC -
SCADA,” in IMECS 2008: International multiconference of
engineers and computer scientists, 2008, vol. II, pp. 1281–1286.
[23] Z. Aydogmus, “Implementation of a fuzzy-based level control
using SCADA,” Expert Syst. Appl., vol. 36, no. 3 part 2, pp.
6593–6597, 2009.
[24] J. C. Vosloo, L. Liebenberg, and D. Velleman, “Case study:
energy savings for a deep-mine water reticulation system,” Appl.
Energy, vol. 92, pp. 328–335, 2012.
[25] A. Nortjé, “DSM strategy for national water pumping systems,”
MEng dissertation, NWU, Potchefstroom, 2012.
[26] N. L. Oosthuizen, “Optimum water distribution between pumping
stations of multiple mine shafts,” MEng dissertation, NWU,
Potchefstroom, 2012.
[27] S. Taljaard, “Comparative study of mine dewatering control
systems,” MEng dissertation, NWU, Potchefstroom, 2018.
[28] Grundfos, “Pump talk,” 2014. [Online]. Available:
http://au.grundfos.com/about-us/news-and-
press/news/PumpTalk.html. [Accessed: 11-Sep-2017].
[29] F. B. Horowitz, B. G. Lipták, and S. Bain, “Pump Controls,” in
Instrument engineers’ handbook, fourth edition, volume two:
process control and optimization, B. G. Lipták, Ed. CRC Press,
2005, pp. 2084–2109.
[30] S. A. Boyer, “SCADA—Supervisory Control and Data
Acquisition,” in Instrument engineers’ handbook, third edition,
process software and digital networks, B. G. Lipták, Ed. CRC
Press, 2002.
[31] B. R. Mehta and Y. J. Reddy, Industrial process automation
systems: design and implementation. Oxford: Butterworth-
Heinemann, 2014.

AUTHORS BIOS AND PHOTOGRAPHS

Stéphan Taljaard holds a Masters
degree in Mechanical Engineering
from the North-West University.

Authorized licensed use limited to: National Institute of Technology Karnataka Surathkal. Downloaded on January 09,2025 at 05:26:36 UTC from IEEE Xplore. Restrictions apply.