A Clustered, Decentralized Approach to Urban Water Management

Round 1

Reviewer 1 Report

## Main comments

The paper reads very well and is structured in an orderly way, especially the Introduction. Using a simplistic approach such as the Euclidian distance is beneficial for planners, as the number of data is decreased and the ease of use is beneficial.
Biggest drawback is the lack of the planning of the pipe network with very crude assumptions of lengths and diameters. Power seems to be only a small problem and differences are only marginal, especially when considering with a yearly energy production of estimated 1600kWh per installed kWp powering pumps.
So I cannot support the conclusion that a clustered approach is beneficial because of less power usage (hence money for power) as the differences are relatively small and can easily be handled. The focus on energy should be shifted to other benefits, maybe maintenance or water security due to a higher number of wells and interconnecting clusters for that.

## Detailed comments

### 2.2

Minimization of distances decreases costs / effort to collect, etc. But what about water security? If the source is contaminated or capacity of a well drops suddenly? And as stated later in the description of the case study: rapid urban growth is likely to happen, which also likely decreases the number of (reliable) sources in the process. In this case (multiple) secure upstream sources are beneficial in terms of water security.

How exactly is leakage minimized with smaller pipes and many small networks? Reference No. 21 does not state that.

### 2.3

Homogeneity may be beneficial in some cases (as stated for re usage of water and recycling). To my knowledge demand variation due to inhomogeneity in an traditional WSS is beneficial as the peak demands of different types of users do not overlap. In an homogeneous system this may be a problem. This should be addressed by the authors.

### 3.3

parcel size in the case study is 150mx150m; Are there any numbers on sensitivity of the model on cluster size, are there any benefits of smaller sizes and increased simulation effort; or drawbacks larger parcels are used? Additionally: a figure with the parcels and population density / water demand would be beneficial (not on a cluster basis).

### 3.4

Q: Isolation of parcels because of the river is not an issue?

Author Response

Reviewer #1 Comments

Main comments

The paper reads very well and is structured in an orderly way, especially the Introduction. Using a simplistic approach such as the Euclidian distance is beneficial for planners, as the number of data is decreased and the ease of use is beneficial.
Biggest drawback is the lack of the planning of the pipe network with very crude assumptions of lengths and diameters. Power seems to be only a small problem and differences are only marginal, especially when considering with a yearly energy production of estimated 1600kWh per installed kWp powering pumps.
So I cannot support the conclusion that a clustered approach is beneficial because of less power usage (hence money for power) as the differences are relatively small and can easily be handled. The focus on energy should be shifted to other benefits, maybe maintenance or water security due to a higher number of wells and interconnecting clusters for that.

Response: For the type of analysis performed herein, it was necessary to simplify the pipe network design, but that does not changes the overall conclusions. The use of a more complex pipe system analysis is considered to be a future research objective. Therefore, we have added a brief paragraph in the discussion that reads: “

While the research documented herein has provides a detailed method that can be used to conduct a cluster analysis, the example application was performed in a simplistic manner to demonstrate the importance of the method. However, more research is required to apply the analysis method developed to more complex utility system design issues with a wider range in pipe diameters and lengths. In addition, the same analysis could be applied to electric utility design, road and street design, and other urban infrastructure. “  

We have added to the discussion of water security as discussed below. However, we do not fully agree with the reviewer concerning a reduction in energy consumption. The key is that within the overall economics of operations, the costs are reduced by a combination of factors including right-sizing of water mains, reduction of operating pressures, and proper placement of ground storage. These factor do have an impact on energy consumption during operation. We have added some text on the topography section that supports the saving of clustering on energy consumption which reads “It is recognized that inhomogeneity of demand pattern (i.e., industrial, commercial, and residential)/variation (evening/day) within a large urban water system can be beneficial as the peak demands of different types of users do not overlap. However, the analysis conducted herein focuses on the homogeneity of source-demand distance, and the reduction of topographic variations within a decentralized cluster to reduce demand pressure variations and effort required to transport water for use/reuse.”

Detailed comments

2.2

Minimization of distances decreases costs / effort to collect, etc. But what about water security? If the source is contaminated or capacity of a well drops suddenly? And as stated later in the description of the case study: rapid urban growth is likely to happen, which also likely decreases the number of (reliable) sources in the process. In this case (multiple) secure upstream sources are beneficial in terms of water security.

Response: With increasing population growth coupled with unsustainable water use practices, cities will face water scarcity due to decrease in reliable and quality water sources. This will impact both locally available and upstream water sources that hydrologically and biologically connected within the catchment. The clustered system’s efficient use of resources is a key reason they are more resilient than conventional urban water systems. Diversification of water sources within the cluster is crucial for achieving water security. In addition to the new section 4.3, text has been added in sections 2.2 and section 3.3

New 2.2 text

The clustered system efficient use of diversified local resources is a key reason why they are more resilient than conventional urban water systems. The sources-demand location plays an important role in reducing the transport distance of water and associated investment cost. Assigning demand to the nearest source location reduces the effort to collect and distribute water to the users. This reduces the cost of the pipe network (due to reduced pipe size/length) required and the energy needed for pumping long distances. Leakage is dependent on system pressure. Basically, the higher the pressure, the larger the leak flow and vice versa [21]. The break rates of water mains depend of several factors such as pipe material, pipe condition and pipe length [22]. The reduction of source-demand distance using a clustered approach results in lesser transport of water for use and reuse, thereby maximizing resource conservation and minimizing losses (i.e. leakage). In addition, it improves the potential to reuse and recycle wastewater to the proximity within the cluster.

New 3.3 Text

In this case study, 10 groundwater sources and 4 potential surface-water abstraction locations were identified (Figure 7). Once the capacity and locations of available sources were identified, the water sources were merged into groups such that the distance between grouped sources was minimized. Since the demand variation is small for the case study of Arua in Uganda, the area was discretized into small parcels of the size 150 m by 150 m. While a smaller/larger parcel size may offer different benefits/drawbacks in the overall analysis, the main principle of decentralized clustered approach is not constrained by the parcel sizes [33] and the sensitivity analysis of parcel resolution is not the focus of this paper. The available water-source abstraction locations of the area were aggregated into seven groups. The decision to propose a number of groups might depend on the size of the area, the size of clusters required, the numbers of water source locations available, etc. Different researchers have highlighted the need for case-by-case analysis to determine the population number that should be supplied by a single source to determine the smaller cluster size [8, 25]. However, the determination of the number of groups required is not the focus of this paper. Thus, the minimum cluster size with a population of 10,000 was used in decentralizing the emerging area as suggested by Webster [5] to determine the number of source centers for grouping. The evaluation of the distance between sources was preformed using Eq. (2). The output of source-group identification process is shown in Figure 7(a) and (b). Once the groups were identified the X, Y coordinate and supply capacity Qs were used to calculate source-centers. The source and source-center information is summarized in Table 1.

New 4.3 Text

4.3. Water Security and Resiliency

Clustered water systems with small ecological footprints that typically use locally available water sources, tend to increase urban water system security. Efficient use of resources is a key reason that clustered systems are more resilient than conventional urban water systems [36,37]. Diversification of water sources is crucial for achieving water security and resiliency. In particular, clustered WSS utilizes non-conventional water sources and fit-for-purpose water supplies. For example, stormwater, grey water and black water can be collected and used close to the source. This provides an additional portfolio of water sources that are not leveraged in existing centralized systems, but that can be better incorporated into decentralized systems and ultimately, improves water security. In addition, clustered systems reduce the amount of water extracted from a centralized water source which usually requires long-distance transport. The diversification of water source allows improvement of environmental flows that are critical for restoring downstream ecosystems in riverine systems.

Clustered systems involve small and modular units that can absorb a high level of disturbances and self-directs the path of adaptation toward a more desirable state [38]. When one cluster system faces a major disturbance, it maintains the capacity of other clusters to perform essential functions. For example, WSS are frequently exposed to leakage and waterborne outbreaks leading to a disruption in the operation; this disruption could cause a water outage across the whole system in a convention utility system. However, clustered systems provide resiliency that allows most areas to operate properly while one cluster is under maintenance. By partitioning the area into separated WSS, the risk of a total system disruption can be reduced. It allows decentralized technologies to be installed within clusters that maintains public health and environmental quality.

How exactly is leakage minimized with smaller pipes and many small networks? Reference No. 21 does not state that.

Response: Leakage is dependent on the system pressure. Basically, the higher the pressure, the larger the leak flow and vice versa. The proposed clustered approach reduces the pressure difference between maximum and minimum elevation points with the cluster approach and the source-demand distance that that results in an efficient, and lesser transport of water for use and reuse. The break rates of water mains depend on several factors such as pipe material, pipe condition and pipe length. The reduction of source-demand distance using clustered approach result lesser transport of water for use and reuse and minimize leakage.

The text in section 2.2 has been modified to reflect this. In addition, reference [21] is replaced, and one more reference [##] is added in the section. 

There are the references

[22] Wang, Y., Zayed, T., & Moselhi, O. (2009). Prediction models for annual break rates of water mains. Journal of Performance of Constructed Facilities, 23(1), 47-54.

2.3

Homogeneity may be beneficial in some cases (as stated for re usage of water and recycling). To my knowledge demand variation due to inhomogeneity in an traditional WSS is beneficial as the peak demands of different types of users do not overlap. In an homogeneous system this may be a problem. This should be addressed by the authors.

Response: Authors agree with the reviewer’s comment. Additional sentences were added on pg 4 to recognize and reflect this perspective.

3.3

Parcel size in the case study is 150mx150m; Are there any numbers on sensitivity of the model on cluster size, are there any benefits of smaller sizes and increased simulation effort; or drawbacks larger parcels are used? Additionally: a figure with the parcels and population density / water demand would be beneficial (not on a cluster basis).

Response: The text in section 3.3 is modified with new sentences to reflect this valuable aspect along with a new reference [36]  

3.4

Q: Isolation of parcels because of the river is not an issue?

Response: We are not sure what the reviewer issue is regarding the isolation of clusters from the river.

Reviewer #2 Comments

Comments and Suggestions for Authors

A much needed research for a real challenge that will occupy the development agencies in the near future.

Response: We thank the reviewer for this comment.

My major suggestions would be to expand the discussion on the tools used for the analysis (K-means; Euclidean norm) and the reasons why these tools were chosen compared to other ones (pros and cons).

Response: More discussion about Euclidean norm was added at the end of section 2.3 with two additional references.  Regarding K-means, the detailed and expanded version of discussion (more than 1 page length) was already included under section 2.8 in the original manuscript, explaining what it is, the origin of the name, why it was chosen over other methods (pros and cons), etc. It is assumed that the reviewer provided this comment because only one short sentence was included at the end of section 2.3, before getting into the section 2.8.

Text added at the end of section 2.3: “  Euclidean norm is chosen as a method of minimization because, in many scientific and engineering applications, the data space is Euclidean and the Euclidean metric has the advantage of being isotropic (rotation invariant) where all vectors are processed in the same way, regardless of their orientation. [23,24].”

Furthermore, when it comes to choosing the resource, there is no discussion on the amount assumed to be provided by the resource. One needs to ensure the sustainable use of the resource. For example, when the groundwater resources were identified, the amount available should be approximately equal to the recharge received by the aquifer. If not, there will be issues of mining the resource and sustainability. Similarly for the surface water resources. If this is not considered by the study, it needs to have a paragraph or two on discussing why.

Response: A full paragraph is added on pg 15 to address this comment. The text reads as follows: “Although, a full water balance analysis (i.e. source availability, recharge rate, current and future demand, water usage from different sectors) for Arua, Uganda is not presented, it should be noted that the sustainable use of the resource needs to be ensured. For example, when different water resources are identified for each decentralized region, the available water sources (ex. groundwater, surface water) and recharge rate from aquifer/infiltration or rainfall can meet the current and future demand of each decentralized region. If not, there will be issues of resource depletion and lack of sustainability.”

There are repetitions in the document that need to be eliminated in the next version of the paper. There are paragraphs that appear identically 4 to 5 times! 

Response: We have attempted to find these repetitions and eliminate some of them.  For example, the whole sentence toward the end of Introduction section was deleted since the nearly same sentence is included in Section 2.1.

Finally, there is a need to include suggestions for future research.

Response: We have added a brief section in the discussion on this issue. This paragraph reads:

 4.4. Future Research

While the research documented herein has provides a detailed method that can be used to conduct a cluster analysis, the example application was performed in a simplistic manner to demonstrate the importance of the method. However, more research is required to apply the analysis method developed to more complex utility system design issues with a wider range in pipe diameters and lengths. In addition, the same analysis could be applied to electric utility design, road and street design, and other urban infrastructure.   

Reviewer 2 Report

A much needed research for a real challenge that will occupy the development agencies in the near future.

My major suggestions would be to expand the discussion on the tools used for the analysis (K-means; Euclidean norm) and the reasons why these tools were chosen compared to other ones (pros and cons).

Furthermore, when it comes to choosing the resource, there is no discussion on the amount assumed to be provided by the resource. One needs to ensure the sustainable use of the resource. For example, when the groundwater resources were identified, the amount available should be approximately equal to the recharge received by the aquifer. If not, there will be issues of mining the resource and sustainability. Similarly for the surface water resources. If this is not considered by the study, it needs to have a paragraph or two on discussing why.

There are repetitions in the document that need to be eliminated in the next version of the paper. There are paragraphs that appear identically 4 to 5 times!

Finally, there is a need to include suggestions for future research.

Author Response

Reviewer #1 Comments

Main comments

The paper reads very well and is structured in an orderly way, especially the Introduction. Using a simplistic approach such as the Euclidian distance is beneficial for planners, as the number of data is decreased and the ease of use is beneficial.
Biggest drawback is the lack of the planning of the pipe network with very crude assumptions of lengths and diameters. Power seems to be only a small problem and differences are only marginal, especially when considering with a yearly energy production of estimated 1600kWh per installed kWp powering pumps.
So I cannot support the conclusion that a clustered approach is beneficial because of less power usage (hence money for power) as the differences are relatively small and can easily be handled. The focus on energy should be shifted to other benefits, maybe maintenance or water security due to a higher number of wells and interconnecting clusters for that.

Response: For the type of analysis performed herein, it was necessary to simplify the pipe network design, but that does not changes the overall conclusions. The use of a more complex pipe system analysis is considered to be a future research objective. Therefore, we have added a brief paragraph in the discussion that reads: “

While the research documented herein has provides a detailed method that can be used to conduct a cluster analysis, the example application was performed in a simplistic manner to demonstrate the importance of the method. However, more research is required to apply the analysis method developed to more complex utility system design issues with a wider range in pipe diameters and lengths. In addition, the same analysis could be applied to electric utility design, road and street design, and other urban infrastructure. “  

We have added to the discussion of water security as discussed below. However, we do not fully agree with the reviewer concerning a reduction in energy consumption. The key is that within the overall economics of operations, the costs are reduced by a combination of factors including right-sizing of water mains, reduction of operating pressures, and proper placement of ground storage. These factor do have an impact on energy consumption during operation. We have added some text on the topography section that supports the saving of clustering on energy consumption which reads “It is recognized that inhomogeneity of demand pattern (i.e., industrial, commercial, and residential)/variation (evening/day) within a large urban water system can be beneficial as the peak demands of different types of users do not overlap. However, the analysis conducted herein focuses on the homogeneity of source-demand distance, and the reduction of topographic variations within a decentralized cluster to reduce demand pressure variations and effort required to transport water for use/reuse.”

Detailed comments

2.2

Minimization of distances decreases costs / effort to collect, etc. But what about water security? If the source is contaminated or capacity of a well drops suddenly? And as stated later in the description of the case study: rapid urban growth is likely to happen, which also likely decreases the number of (reliable) sources in the process. In this case (multiple) secure upstream sources are beneficial in terms of water security.

Response: With increasing population growth coupled with unsustainable water use practices, cities will face water scarcity due to decrease in reliable and quality water sources. This will impact both locally available and upstream water sources that hydrologically and biologically connected within the catchment. The clustered system’s efficient use of resources is a key reason they are more resilient than conventional urban water systems. Diversification of water sources within the cluster is crucial for achieving water security. In addition to the new section 4.3, text has been added in sections 2.2 and section 3.3

New 2.2 text

The clustered system efficient use of diversified local resources is a key reason why they are more resilient than conventional urban water systems. The sources-demand location plays an important role in reducing the transport distance of water and associated investment cost. Assigning demand to the nearest source location reduces the effort to collect and distribute water to the users. This reduces the cost of the pipe network (due to reduced pipe size/length) required and the energy needed for pumping long distances. Leakage is dependent on system pressure. Basically, the higher the pressure, the larger the leak flow and vice versa [21]. The break rates of water mains depend of several factors such as pipe material, pipe condition and pipe length [22]. The reduction of source-demand distance using a clustered approach results in lesser transport of water for use and reuse, thereby maximizing resource conservation and minimizing losses (i.e. leakage). In addition, it improves the potential to reuse and recycle wastewater to the proximity within the cluster.

New 3.3 Text

In this case study, 10 groundwater sources and 4 potential surface-water abstraction locations were identified (Figure 7). Once the capacity and locations of available sources were identified, the water sources were merged into groups such that the distance between grouped sources was minimized. Since the demand variation is small for the case study of Arua in Uganda, the area was discretized into small parcels of the size 150 m by 150 m. While a smaller/larger parcel size may offer different benefits/drawbacks in the overall analysis, the main principle of decentralized clustered approach is not constrained by the parcel sizes [33] and the sensitivity analysis of parcel resolution is not the focus of this paper. The available water-source abstraction locations of the area were aggregated into seven groups. The decision to propose a number of groups might depend on the size of the area, the size of clusters required, the numbers of water source locations available, etc. Different researchers have highlighted the need for case-by-case analysis to determine the population number that should be supplied by a single source to determine the smaller cluster size [8, 25]. However, the determination of the number of groups required is not the focus of this paper. Thus, the minimum cluster size with a population of 10,000 was used in decentralizing the emerging area as suggested by Webster [5] to determine the number of source centers for grouping. The evaluation of the distance between sources was preformed using Eq. (2). The output of source-group identification process is shown in Figure 7(a) and (b). Once the groups were identified the X, Y coordinate and supply capacity Qs were used to calculate source-centers. The source and source-center information is summarized in Table 1.

New 4.3 Text

4.3. Water Security and Resiliency

Clustered water systems with small ecological footprints that typically use locally available water sources, tend to increase urban water system security. Efficient use of resources is a key reason that clustered systems are more resilient than conventional urban water systems [36,37]. Diversification of water sources is crucial for achieving water security and resiliency. In particular, clustered WSS utilizes non-conventional water sources and fit-for-purpose water supplies. For example, stormwater, grey water and black water can be collected and used close to the source. This provides an additional portfolio of water sources that are not leveraged in existing centralized systems, but that can be better incorporated into decentralized systems and ultimately, improves water security. In addition, clustered systems reduce the amount of water extracted from a centralized water source which usually requires long-distance transport. The diversification of water source allows improvement of environmental flows that are critical for restoring downstream ecosystems in riverine systems.

Clustered systems involve small and modular units that can absorb a high level of disturbances and self-directs the path of adaptation toward a more desirable state [38]. When one cluster system faces a major disturbance, it maintains the capacity of other clusters to perform essential functions. For example, WSS are frequently exposed to leakage and waterborne outbreaks leading to a disruption in the operation; this disruption could cause a water outage across the whole system in a convention utility system. However, clustered systems provide resiliency that allows most areas to operate properly while one cluster is under maintenance. By partitioning the area into separated WSS, the risk of a total system disruption can be reduced. It allows decentralized technologies to be installed within clusters that maintains public health and environmental quality.

How exactly is leakage minimized with smaller pipes and many small networks? Reference No. 21 does not state that.

Response: Leakage is dependent on the system pressure. Basically, the higher the pressure, the larger the leak flow and vice versa. The proposed clustered approach reduces the pressure difference between maximum and minimum elevation points with the cluster approach and the source-demand distance that that results in an efficient, and lesser transport of water for use and reuse. The break rates of water mains depend on several factors such as pipe material, pipe condition and pipe length. The reduction of source-demand distance using clustered approach result lesser transport of water for use and reuse and minimize leakage.

The text in section 2.2 has been modified to reflect this. In addition, reference [21] is replaced, and one more reference [##] is added in the section. 

There are the references

[22] Wang, Y., Zayed, T., & Moselhi, O. (2009). Prediction models for annual break rates of water mains. Journal of Performance of Constructed Facilities, 23(1), 47-54.

2.3

Homogeneity may be beneficial in some cases (as stated for re usage of water and recycling). To my knowledge demand variation due to inhomogeneity in an traditional WSS is beneficial as the peak demands of different types of users do not overlap. In an homogeneous system this may be a problem. This should be addressed by the authors.

Response: Authors agree with the reviewer’s comment. Additional sentences were added on pg 4 to recognize and reflect this perspective.

3.3

Parcel size in the case study is 150mx150m; Are there any numbers on sensitivity of the model on cluster size, are there any benefits of smaller sizes and increased simulation effort; or drawbacks larger parcels are used? Additionally: a figure with the parcels and population density / water demand would be beneficial (not on a cluster basis).

Response: The text in section 3.3 is modified with new sentences to reflect this valuable aspect along with a new reference [36]  

3.4

Q: Isolation of parcels because of the river is not an issue?

Response: We are not sure what the reviewer issue is regarding the isolation of clusters from the river.

Reviewer #2 Comments

Comments and Suggestions for Authors

A much needed research for a real challenge that will occupy the development agencies in the near future.

Response: We thank the reviewer for this comment.

My major suggestions would be to expand the discussion on the tools used for the analysis (K-means; Euclidean norm) and the reasons why these tools were chosen compared to other ones (pros and cons).

Response: More discussion about Euclidean norm was added at the end of section 2.3 with two additional references.  Regarding K-means, the detailed and expanded version of discussion (more than 1 page length) was already included under section 2.8 in the original manuscript, explaining what it is, the origin of the name, why it was chosen over other methods (pros and cons), etc. It is assumed that the reviewer provided this comment because only one short sentence was included at the end of section 2.3, before getting into the section 2.8.

Text added at the end of section 2.3: “  Euclidean norm is chosen as a method of minimization because, in many scientific and engineering applications, the data space is Euclidean and the Euclidean metric has the advantage of being isotropic (rotation invariant) where all vectors are processed in the same way, regardless of their orientation. [23,24].”

Furthermore, when it comes to choosing the resource, there is no discussion on the amount assumed to be provided by the resource. One needs to ensure the sustainable use of the resource. For example, when the groundwater resources were identified, the amount available should be approximately equal to the recharge received by the aquifer. If not, there will be issues of mining the resource and sustainability. Similarly for the surface water resources. If this is not considered by the study, it needs to have a paragraph or two on discussing why.

Response: A full paragraph is added on pg 15 to address this comment. The text reads as follows: “Although, a full water balance analysis (i.e. source availability, recharge rate, current and future demand, water usage from different sectors) for Arua, Uganda is not presented, it should be noted that the sustainable use of the resource needs to be ensured. For example, when different water resources are identified for each decentralized region, the available water sources (ex. groundwater, surface water) and recharge rate from aquifer/infiltration or rainfall can meet the current and future demand of each decentralized region. If not, there will be issues of resource depletion and lack of sustainability.”

There are repetitions in the document that need to be eliminated in the next version of the paper. There are paragraphs that appear identically 4 to 5 times! 

Response: We have attempted to find these repetitions and eliminate some of them.  For example, the whole sentence toward the end of Introduction section was deleted since the nearly same sentence is included in Section 2.1.

Finally, there is a need to include suggestions for future research.

Response: We have added a brief section in the discussion on this issue. This paragraph reads:

 4.4. Future Research

While the research documented herein has provides a detailed method that can be used to conduct a cluster analysis, the example application was performed in a simplistic manner to demonstrate the importance of the method. However, more research is required to apply the analysis method developed to more complex utility system design issues with a wider range in pipe diameters and lengths. In addition, the same analysis could be applied to electric utility design, road and street design, and other urban infrastructure.