مطالعه ی مقایسه ای دو معیار کنترلی مختلف برای مدیریت زمان- واقعی آب بندهای زیرزمینی شهری

We present the comparison of two control criteria for the real-time management of a water well field. The criteria were used to simulate the operation of the Hardhof well field in the city of Zurich, Switzerland. This well field is threatened by diffuse pollution in the subsurface of the surrounding city area. The risk of attracting pollutants is higher if the pumping rates in four horizontal wells are increased, and can be reduced by increasing artificial recharge in several recharge basins and infiltration wells or by modifying the artificial recharge distribution. A three-dimensional finite elements flow model was built for the Hardhof site. The first control criterion used hydraulic head differences (Δh-criterion) to control the management of the well field and the second criterion used a path line method (%s-criterion) to control the percentage of inflowing water from the city area. Both control methods adapt the allocation of artificial recharge (AR) for given pumping rates in time. The simulation results show that (1) historical management decisions were less effective compared to the optimal control according to the two different criteria and (2) the distribution of artificial recharge calculated with the two control criteria also differ from each other with the %s-criterion giving better results compared to the Δh-criterion. The recharge management with the %s-criterion requires a smaller amount of water to be recharged. The ratio between average artificial recharge and average abstraction is 1.7 for the Δh-criterion and 1.5 for the %s-criterion. Both criteria were tested online. The methodologies were extended to a real-time control method using the Ensemble Kalman Filter method for assimilating 87 online available groundwater head measurements to update the model in real-time. The results of the operational implementation are also satisfying in regard of a reduced risk of well contamination.

This paper presents a real-time control methodology using two different control criteria which has been put into practice for the safe abstraction of drinking water in the city of Zurich, Switzerland. The Hardhof well field delivers 15% of the drinking water demand of the city and serves as one of the main waterworks of the urban water supply net. The growth of residential and industrial areas close to the Hardhof well field over recent decades has led to much higher potential contamination risks in the ground and at the surface. The historic basis for well field management relied on defined well head protection zones to avoid contaminating activities close to the wells and the additional monitoring of the pumped water quality. Sporadic tracer studies and collections of water samples were used to locate and delineate contamination sources. Meanwhile the HACCP concept (Hazard Analysis and Critical Control Points) (WHO, 2010) was incorporated as a legal obligation in Switzerland (EDI, 2010). It requires all producers of drinking water to guarantee the quality standard of the supplied water at any time. Therefore, the possible inflow of water from parts of the aquifer which may contain sources of contamination must be monitored or controlled in real-time. Online-sensors can be used for the monitoring of aquifers and the operation of wells. These sensors transfer head data, temperature or chemical data, e.g. electrical conductivity in the groundwater and have potential value for the real-time management of well fields. We consider a real-time well field management system as a combination of three technical parts: 1) real-time transferred data, 2) a model that is updated with this data, and 3) a control algorithm calculating the necessary pumping rates at current time or as a predictive signal for future management decisions. Our current study introduces a control approach which uses path line analysis of particles as control criterion for the real-time management of the Hardhof well field under conditions of temporally variable forcings (natural recharge rates, river stages, boundary conditions and groundwater management).

The %s-criterion is a surrogate control criterion. For each day the path lines are calculated with the instantaneous flow field, assuming that all conditions stay constant. If the %s-criterion is optimally controlled (i.e., %s = 0) for each day of the time period between 1st of January 2004 and 23rd of August 2005, the criterion can be regarded as conservative and robust because its requirement regarding optimality (i.e., %s = 0) does not allow inflow of city water. The limitations of the %s-criterion lie within the maximum travel time for the calculation of the path lines. We chose a maximum of 200 days as abort criterion for the calculation of the path lines which has no implications on the validity of the results (more than 99.5% of the particles reached the defined control planes in x-, y- and z-direction or the defined boundaries of the river or the city domain). In comparison to the path line method, a transport model for the transient simulation of electrical conductivity in the groundwater would be better suited to estimate EC values in the horizontal wells. However, this is far from trivial as the EC of the groundwater is modified during its transport below the city. The space-time distribution of the processes and sources that result in the EC-increase of the city water are largely unknown and therefore it is difficult to simulate these processes in a reliable manner. A dense network of online sensors that measure EC could alleviate this problem, and these measurement data could be assimilated by a transport model to update the spatial distribution of EC in time. So far, the online application in the field showed that the %s-criterion for the control leads to reduced (measured) EC values of the pumped drinking water, indicating a reduced inflow of city water. The CPU-time per optimization iteration is ca. 2.5 min yielding 12.5 min for 5 iterations. This is the main reason to limit the number of particles to 1350. The number of particles could be adapted to 2000 or even 3000 in order to produce better results in terms of resolution. The introduced %s-concept could be used for different classes of problems, mainly for all groundwater model calculations using the finite element approach. Well sites without artificial recharge basins, yet threatened by all kinds of contamination in the soil or groundwater could rely on this concept in order to adapt the necessary abstraction rates in real-time.