Our Services

Some key outputs for the model include the following:

The model is mapped on a Google earth backdrop of the Anglo’s Rustenburg Operation with mining infrastructure located at their actual geographical sites.  It is therefore not an unreferenced static line diagram which is difficult to relate to.

The model runs in hourly time steps adding crucial sensitivity analyses of events that may take place in the framework of only a few hours.  Examples are 1:50 year storm events that may occur in a matter of a few hours, unscheduled maintenance events and overflows to the environment which can have significant consequences in a matter of hours.  Not taking hours in consideration is where we often miss it in the mining sector.  The opportunity is often missed because we are dealing with pumps running at cubs per hour, but dams filling at cubs per day.  Therefore, to simply argue that that short times steps do not add any value at operational level over a period of 20 years is not a simple linear assumption to be made.  Most plants work with planted and unplanned maintenance that last several hours rather than days and this needs to be factored in the model and it is easier to handle when using hours, if days are used it is mismatched.  Further to this, when engaging the site you need to have a model available which the ops team can related to.  For example, a 37 kW pump at the SWP may pump (considering the pipe and diameter) at 100 cubs per hour but over a period of 1 day or 30 days the average yield might be 65cubs per month of 87cubs per day.  Therefore, reporting back at high level steps of 1 day and even monthly may be good enough, but feeding back to site and addressing the issue it is found that hourly speaks more directly to the guys at coalface.

The model integrates all dynamic variables that impact storage levels.  Variable rainfall intensity, variable pool sizes of dams, seepage and evaporation rates, production rates and flow logic are all determined dynamically during a simulation run.

The model integrates current operations with future expansion plans.

The model records the frequencies and severity of droughts and storm events to show how often dams will run empty or with what volumes and when it will overflow.

The model calculates the average water requirements over life of operations.  The model also identify shortfalls for days and months to be expected.  The averages is simply ho DWS wants in and is also an output.

The model identifies risks or bottlenecks directly from the output graphs.

The model allows for scenario analysis.

The model uses a rainfall generator that dynamically simulates future rainfall based on historical values (Exceedance probabilities comments).  The water balance model is built with various random distribution functions that generate events with a natural spread in variance.  For example:

Maintenance schedules has a natural mean time to failure (MTTF) and mean time to repair (MTTR) that uses an exponential distribution to estimate the time it will take before the next unscheduled maintenance event will occur.  The repair time also uses an exponential distribution to estimate the time lapse before that component is functional again.  Plants, shafts, boreholes and any other significant components in the model may use these functions (each using their own unique MTTF and MTTR average values) to simulate natural unscheduled maintenance events.  Scheduled maintenance may be added if it is known.

Rainfall events similarly use other random distribution functions to estimate the time between rainfall events, the duration of each rainfall event and the intensity of each rainfall event (a combination of Normal-, Exponential- and Poison distributions are used).  These allow for the natural spread of rainfall events and their intensities for different months of the year and the end results are calibrated to ensure long-term monthly average rainfall is adhered to while also producing the natural 1:20 or 1:50 storm events that might occur.

The blend of the distribution functions above and the operational logic applied for each flow (when to switch pumps on/off) results in very realistic simulated results of what “might” happen for any given simulated period.  To ensure that a given simulation run does not produce a skewed result set we use stochastic modelling techniques whereby many repetitions are run for the same simulation period.  Some runs produce below average rainfall and the next may produce above average values but given enough repetitions it will start to average out.  Using statistical analysis techniques, we can then derive the exceedance probabilities that for example will state that “we have a 95% confidence level that the storm water dam will not overflow more than once in a simulated fifty year period”.

Testing various water conservation- and water demand management (WC/WDM) strategies will therefor allow combined Operations to make informed decisions as to how much each strategy will impact on the critical parameters (like overflows to the environment or how much water will be required externally during peak production periods).

The Water Hunters model allows the user to see the current and forecasted levels of all dams and flow volumes over the full simulation period by means of user-friendly Excel result files.  Results are presented in graphical and numerical format for easy analysis by the user.

A static water balance provides a summary of gains and losses for any period selected by the user.  This is a valuable tool that allows a snapshot of the complete project in one glance.