ArticlePDF Available

Desalination network model driven decision support system: A case study of Saudi Arabia


Abstract and Figures

This study aims to develop a network model driven platform that supports decision-makers to make well-informed decisions for the efficient water supplies, taking Saudi Arabia as a case study. The water/energy network analysis should be able to identify optimal locations for sustainable desalination infrastructure investments, accounting for the existing assets and the current investment plans. The geographical aspect of individual resource production and distribution can be quantitatively handled by a graph-theoretic approach. This study employs a new multicommodity network flow model called the INFINIT (interdependent network flows with induced internal transformation) model, which enables to address water-energy nexus issues and to optimize the flow of multiple resources as well as placement of new water/energy facilities at the individual facility level. The INFINIT model in this study formulates and solves mixed-integer linear programming (MILP) problems to minimize the designated multi-objective functions of the total cost and CO2 emission. As a result of optimization, the Pareto-optimal solutions with different network flow topology and the downselected potential locations for new facilities are obtained. To effectively visualize alternative design and policy scenarios, two ways of visualization of the results are developed: a MATLAB-based graphical user interface and tabletop 3D map projection.
Content may be subject to copyright.
Desalination network model driven decision support system:
a case study of Saudi Arabia
Takuto Ishimatsu a,*, Abdelkrim Doufene a, Abdullah Alawad b, Olivier de Weck a
a Massachusetts Institute of Technology, USA
b King Abdulaziz City for Science and Technology, Saudi Arabia
This study aims to develop a network model driven platform that supports decision-makers to make well-
informed decisions for the efficient water supplies, taking Saudi Arabia as a case study. The water/energy
network analysis should be able to identify optimal locations for sustainable desalination infrastructure
investments, accounting for the existing assets and the current investment plans. The geographical aspect of
individual resource production and distribution can be quantitatively handled by a graph-theoretic approach.
This study employs a new multicommodity network flow model called the INFINIT (interdependent network
flows with induced internal transformation) model, which enables to address water-energy nexus issues and
to optimize the flow of multiple resources as well as placement of new water/energy facilities at the individual
facility level. The INFINIT model in this study formulates and solves mixed-integer linear programming
(MILP) problems to minimize the designated multi-objective functions of the total cost and CO2 emission. As
a result of optimization, the Pareto-optimal solutions with different network flow topology and the
downselected potential locations for new facilities are obtained. To effectively visualize alternative design and
policy scenarios, two ways of visualization of the results are developed: a MATLAB-based graphical user
interface and tabletop 3D map projection.
Keywords: desalination network; decision support system; multicommodity flow; optimization modeling
History: Presented at IDA World Congress 2015 August 2015; Submitted for journal publication March 2017
1. Introduction
The Kingdom of Saudi Arabia (KSA) is at the forefront of water desalination for human use. This is driven
by the scarcity of renewable water resources. Indeed, with limited rainfall and excessive consumption, the
major ground water aquifers are being depleted. A study in 1984 estimated that the storage of the main and
secondary aquifers was 500 billion cubic meters and a study in 1996 estimated the amount to be 289.1 billion
cubic meters [1]. Considering the reported consumption rates since then, the state of ground water resources
in KSA is becoming unsustainable. Thus, water desalination in KSA is an indispensable strategic choice to
secure potable water. Desalination is an energy intensive process. The energy required by those desalination
plants are mainly supplied by oil and gas. Today, there is no use of nuclear energy or coal, and there is very
limited use of renewable energy.
Growing demand on both water and energy due to the rapid population growth puts significant pressure
on the current existing infrastructure [2]. This situation may lead to major problems for KSA if not tackled in
an optimal manner. Drastic measures should be taken from either demand or supply side to put the state of
water resources in a more sustainable path. On the supply side, large investments will be required in new
water desalination and power generation plants and their accompanying infrastructure such as transmission
and distribution systems. Planning a new infrastructure requires a better understanding of the current
network of infrastructure as well as the dynamics of demand. Since infrastructure projects are usually
expensive, decisions must be made with a full understanding of the effect of a new project on the overall
infrastructure network.
This study aims to develop a network model based platform that supports decision-makers to make well-
informed near-term and long-term decisions at a policy level for the efficient water supplies. While single
facility analysis is very informative, the drive to use renewables as a primary energy source for desalination
will likely impact the choice of both water and energy facility locations. The water/energy network model views
a single plant as part of a larger network representing a region or spanning the entire KSA. Since desalination
plants and power plants are increasingly connected together with water pipelines and electricity transmission
lines, evaluating a standalone plant as a node in a larger network is compelling. The water/energy network
analysis should be able to investigate and identify candidate locations for sustainable desalination
infrastructure investments, accounting for the existing assets and the current investment plans.
In addition to the network model, due to the high degree of complexity of the problems being addressed, it
is also important that the decision support system (DSS) is able to effectively visualize alternative design and
policy scenarios. Fig. 1 illustrates the DSS proposed in this paper. Two ways of visualization of the results are
developed: the MATLAB-based graphical user interface (GUI) and the tabletop 3D map projection. The DSS
also includes a common database that covers all water and energy data needed for both supply and demand
sides. This network model driven DSS will enable stakeholders to both evaluate and refine complex scenarios
addressing the location of water and energy investments across KSA.
Fig. 1. Network model driven decision support system.
2. Related work
While a great deal of work has been done on optimizing the unit operations of a single desalination plant,
little attention has been paid to optimizing the whole water supply chain network. Optimization of water
supply chain considers higher level of strategic decisions about the optimal water supply and distribution to
meet geographically dispersed demand nationwide. Kondili et al. proposed a linear programming optimization
model for the optimal planning of complex water systems with multiple supply sources (desalination, ground
reservoirs, dams, and water transfer) and multiple user demands (agriculture, industry, and urban and other
sectors) [3]. However, this model only looks at the optimal matching between these sources and users without
addressing a geographical network and its constraints. Al-Nory et al. proposed a mixed integer linear
programming model to solve a water desalination supply chain problem as a network flow problem to provide
decision makers with a set of investment alternatives comprising combinations of different desalination plant
locations, capacities, technologies, and energy sources [4,5]. In this model, however, a simplified network with
limited numbers of nodes and arcs was assumed in the analysis so that only obvious matching of the supply
and demand locations occurred due to a limited tradespace. In addition, constraints on water transmission
such as pumping energy and water losses (evaporation and leakage) are not given or only given through a unit
flow cost. Also in this model, water-energy nexus issues cannot be addressed because of the decoupling of water
layer from energy layer.
3. Water/energy network model
The water/energy network model discussed here views a single desalination/power plant as part of a larger
network representing a region or the entire KSA. Since desalination/power plants are increasingly connected
together with water pipelines and electricity transmission lines, the impact of evaluating a standalone plant
versus evaluating that same plant as one of the nodes in a larger network will be interesting.
The geographical aspect of individual resource production and distribution can be quantitatively handled
by a graph-theoretic approach. As shown in Fig. 2, since multiple interacting resources or “commodities” share
the same network such as water (potable, non-potable, and waste water) and electricity, this problem can be
modeled as a multi-commodity network flow problem. In this context, a new network flow modeling method
was developed to optimize the flow of resources and placement of new facilities (and expansion or retirement
of existing facilities) at the individual facility level [6,7]. This method, which is called the INFINIT
(Interdependent Network Flows with Induced Internal Transformation) method, has demonstrated successful
applications in different contexts [8,9].
Fig. 2. KSA multi-dimensional infrastructure network.
Before discussing the context-specific formulation, we present the basic form of the INFINIT model. The
INFINIT model was created by combining two forms of generalization of the classical minimum cost flow
problem: generalized flow problems, in which arcs might consume or generate flows, and multicommodity flow
problems, in which multiple commodities share the same network. In INFINIT, the flow  in arc  is split
into two parts: 
and 
, where 
represents the outflow from node and 
 represents the inflow into node
. Let 
and 
 denote the cost per unit outflow and inflow, respectively. Using these notations, the INFINIT
problem can be formulated as follows:
  
subject to:
 
  
    
 
 
 
 
   
 
 
 
 
 
where denotes the net supply/demand vector at node , 
denotes constants associated with arc , and
and 
represent the lower bound and the capacity of arc , respectively. There are three types of matrix
multiplication introduced in Eqs. (2)-(4): 
is called a
flow equilibrium matrix
,  is called a
transformation matrix
, and 
is called a
flow concurrency matrix
. With this modification, we can treat the
flow gain/loss due to the interaction between commodities and the flow transformation between commodities.
Fig. 3 illustrates a minimum network containing necessary components in the KSA water/energy system.
In the INFINIT model, resource processing is also modeled as a loop (instead of a node) that transforms one
commodity to another at the same node. A loop is an arc that connects a node to itself, but mathematically can
be handled in the same manner as a normal arc. Supply nodes include desalination plants (blue) and power
plants (yellow). Desalination/power plants can provide both potable water and electricity. Demand nodes are
cities (red), some of which have waste water treatment capability (green nodes with loops). Arcs represent
water pipelines or electricity transmission lines while loops represent local water production at desalination
plants, electricity generation at power plants, or waste water treatment in cities that have such facilities. In
the scope of this study, a flow equilibrium matrix 
models power consumption for water processing (loops)
and pumping energy (arcs), and a flow transformation matrix  models water processing/electricity
generation (loops) and water/electricity loss during transmission (arcs). A flow concurrency constraint in Eq.
(4) represents nothing in this context and therefore we can ignore it by setting 
to be zero.
Fig. 3. INFINIT minimum network containing necessary components.
While the optimality of flow can be defined by the objective function in many ways, it should be something
that reflects the sustainability of the infrastructure systems such as the total cost (CAPEX and OPEX) and
the total CO2 emission. The social issue related to water shortages is enforced by the supply-demand constraint
in Eq. (2).
Mixed-integer linear programming formulation
This subsection presents the context-specific formulation using mixed-integer linear programming (MILP).
One of the goals of solving the INFINIT problem in this context is to explore the best investment strategy in
future infrastructure systems and we wish to determine where we should build a new facility and how much
capacity it should have in the context of the overall system. This decision-making can be included in the
problem by introducing another variable, 
, representing the capacity of a newly established facility. We
must also consider fixed costs. For example, if we actively use a pipeline, then operations and maintenance
cost is incurred regardless of how much water is transported through it. In contrast to the proportional
variable costs, this cost can be modeled as a fixed cost, which is incurred if we use the arc and not incurred if
we do not. For this reason, we also introduce a binary variable, , representing whether a particular arc in
the system is used or not. Using these notations, the INFINIT MILP formulation for the context described in
this paper is as follows:
  
subject to Eqs. (2)-(3) and:
 
 
 
 
   
 
    
  
The objective function in Eq. (6) includes three types of costs: costs proportional to the flow 
, costs
proportional to the capacity expansion 
, and fixed costs incurred by . Since the coefficients can be defined
differently for individual arcs/loops (facilities), the location-specific cost can also be taken into account. For
example, if land cost at a particular node has a capacity-dependent part and a fixed part, it can be modeled
by tuning the 
and 
coefficients. The flow bound constraint in Eq. (7) is now involved the capacity
expansion 
as well as the existing capacity 
. For a potential arc (facility), 
  so that the flow must be
less than or equal to its design capacity 
. The constraint in Eq. (8) is the so-called big- method. is
chosen sufficiently large. This constraint forces    whenever 
  for any commodity. If, on the other
hand, we do not use arc , then   , which forces 
 . The last constraint in Eq. (9) simply states
that  is binary.
As shown in Fig. 4, multiple loops can be defined associated with a single node. For example, when we
design a new desalination plant, we have different technology options such as RO, MSF, and MED. Likewise,
for a power plant, we have several options such as conventional natural gas and diesel, PV, and CSP. Each
technology has different parameters (energy consumption, CAPEX, OPEX, CO2 emission, etc.), which are
reflected in , 
, and . Optimization will automatically determine which loop (technology) should be built
(employed) at node .
Fig. 4. A single node with multiple loops representing technology options.
INFINIT MILP model structure
The INFINIT MILP model has three phases. Fig. 5 shows the model structure. The first phase is a common
database. The input parameters and assumed values used in this analysis are listed in Table 3 in Appendix.
This database has a list of major cities, major desalination plants and power plants, water pipeline networks,
and electricity transmission line networks. Figures 12-14 present the lists of cities, desalination plants, and
power plants used in this analysis, respectively. For each entry, it includes information such as GPS
coordinates, city population, plant capacity, pipeline/powerline distance, capacity expansion cost, operation
cost, etc. This list defines a set of existing nodes and arcs. In addition to them, placeholder nodes and arcs
for potential new infrastructure need to be listed in the database, which requires at least GPS coordinates of
the candidate sites. If there is a location-specific consideration (e.g., land cost), it should be included in the
database. Technology and capacity selection is what the optimizer does. Note that optimization explores not
only newly constructed facilities but also capacity expansion of the existing facilities. The second phase is the
preprocessing phase. Before solving an optimization problem, we need to generate a problem itself by
translating the database into a network flow optimization problem. In this phase, the Excel spreadsheet is
read into MATLAB and the INFINIT MILP parameters in the above equations are constructed. As a by-
product of the preprocessing phase, we obtain a network graph (lower middle in Fig. 5) on which we will solve
a network flow problem. In the third phase, we set the weights between multi-objective functions and solve
the problem using IBM CPLEX Optimizer within the MATLAB environment. Finally we obtain numerical
results as well as a network graph of optimized flow (lower right in Fig. 5).
Fig. 5. INFINIT MILP model structure.
Regarding the computation time, it is obvious that the MILP optimization takes some time but also the
preprocessing phase is not a short process. The network has approximately 200 nodes and 800 arcs and a huge
matrix (approx. 20000x20000) must be generated to define a problem. Using MATLAB 8.3 (R2014a) on an
Intel® CORETM i7-2640M CPU at 2.80 GHz, the preprocessing time of the current version is about 10 seconds.
For the optimization phase, since a MILP problem of this size is not completely solvable, we need to define
some acceptable optimality gap to quit the calculation within a reasonable time. In most cases, the optimality
gap drops down to less than 3% within a minute.
Example results
As an example, the result of the application of INFINIT MILP for the KSA water system is shown below.
Figure 6 shows a network graph, including 97 city nodes in red and green, 42 desalination plants in blue (some
are aggregate nodes combining multiple desalination plants at the same geographic location into one), and 69
power plants, which are online or presumed online as of 2010. The size of each node is proportional to the city
population or plant capacity. The blue lines are the working pipelines of the Saline Water Conversion
Corporation (SWCC) as of 2010. The dotted lines represent potential connections that do not yet exist but are
considered as candidates in the optimization.
The decision variable vector used in this analysis is defined as:
feed water
potable water
non-potable water
waste water
energy source
where the first four variables are related to water and the last two are related to electricity. A capacity
expansion vector 
and a decision binary vector  are also defined in the same way. The objective function
is established by combining the total cost (CAPEX and OPEX) and the CO2 emission using a weight and
rescaling factors. In this paper, a transportation layer is not considered for the following two reasons. First,
the contribution of transportation layer (road, rail, water, and air) in the water/electricity transmission and
distribution phase is negligibly small. Second, the KSA transportation layer is so large and complex (347 nodes
and 1578 arcs) that adding it to the model would significantly slow down the computation for the sake of
negligible benefit.
In defining supply and demand, we must first determine a basis (initial) year and a target (final) year as
inputs to the model. A basis year represents the “current” year that the infrastructure capacity is based on
and a target year represents the “future” year to which the water demand corresponds. Here a basis year and
a target year are set to 2010 and 2030, respectively. The problem is to optimize both the investment and
operations of the system to meet the expected demand in 2030 using the 2010 supply capability as a starting
point. The total population in KSA in 2030 is expected to be about 42 million, which is about 1.5 times that of
2010. As shown in Fig. 6, 97 cities are included in the optimization problem, covering about 80% of the total
population in KSA. We assume this coverage remains the same in 2030 and therefore the total demand for
water in this problem is 8.03 million cubic meters per day (MCM/day). Of this, 5.62 MCM/day can only be
satisfied with potable water while the remaining 2.41 MCM/day can be satisfied with either potable or non-
potable water (recycled waste water). Meanwhile, the 2010 desalination capacity totals 3.27 MCM/day and the
ground water capacity totals 2.29 MCM/day. The 2010 capacity falls short of the potable water demand in
2030 even without taking geography into consideration. Thus it is inevitable to expand the capacity of the
existing infrastructure, constructing new infrastructure, or changing per-capita demand or a combination of
these. One of the objectives of the INFINIT network model is to find the optimal facility-level strategy to meet
future demand in consideration of unique geographical features within KSA. Results are shown in Table 1 and
Fig. 7.
Table 1 lists the results with the leftmost column representing the weighting factor between total cost and
CO2 emissions, and Fig. 7 plots the corresponding Pareto front. Note that CAPEX is divided equally across the
lifetime of a facility, which is assumed to be 25 years unless otherwise specified. We observe that as the weight
of the total cost in the objective function decreases, the total cost of the KSA water system gradually increases
from 6.68 to 11.00 billion USD per year (BUSD/year) and the CO2 emissions decrease from 17.82 to 11.99
million metric tons per year (Mmt/year). The total water production ranges between 2,474 and 2,130
MCM/year and the total electricity generation ranges between 18,445 and 20,070 GWh/year. In Figure 7, since
none of these 11 points are dominated, they all form a Pareto-optimal set.
Fig. 6. Network graph with 208 nodes and 812 arcs (including 208 loops).
Table 1
List of INFINIT optimization results (Pareto front).
Total Cost
Total Water
Fig. 7. CO2 emission vs. total cost for Pareto-optimal 2030 KSA water infrastructure.
As a representative example, Fig. 8 shows the resulting network for a weight of 0.5, that is, equal weight
on the total cost and CO2 emission (corresponding to the point within the light red highlighted circle in Fig.
7). The size of a desalination plant node or a power plant node represents the amount of water produced or
electricity generated in the resulting network flow. Blue lines are the pipelines that are in use and they are
not necessarily the existing ones (in 2010). For example, we can observe that Medina serves as a hub for
supplying water up to Hail and several cities in the central (Al-Qassim) region through long-distance pipelines.
Since cities in the central region are currently connected from the east coast via Riyadh, it is interesting that
the resulting network has chosen the supply from the west coast. This implies the possibility of a future trans-
peninsula pipeline network connecting the east and west coasts. An additional benefit, not represented in the
current objective function, is the increased robustness which comes from sourcing water both from the Red
Sea and the Gulf.
Table 2 lists the top 10 CAPEX investment projects for this particular KSA water infrastructure portfolio.
The largest CAPEX investment is the construction of a massive SWRO desalination plant at Ras Al-Khair,
which is represented by the biggest blue node in the Eastern region. The new capacity is about 2.51 MCM/day,
which is almost 2.5 times that of the existing plant as of 2017. The next 4 projects (top 2 through 5) fall into
the category of capacity expansion of several existing desalination plants. Projects 6 through 9 are the
construction of new long-distance pipelines. The pipeline from Ras Al-Khair to Riyadh already exists as of
2017, while the remaining three are suggested as new projects in this INFINIT optimization result. The last
one is the construction of a power plant at Ras Al-Khair, whose capacity is about 793 [MW]. Under the current
technology choices, this is envisioned as a gas-fired power plant, but it could conceivably also be implemented
using solar or nuclear technology. This result supports the geographical advantage of Ras Al-Khair, which has
actually been constructed and in full commercial operation since 2016.
Fig. 8. Resulting network graph for 2030 KSA water infrastructure: equal weight for total cost (50%) and
CO2 emission (50%).
Table 2
Top 10 CAPEX investment purposes for 2030.
Investment Purpose
Desal plant construction
Ras Al-Khair
2512300 [m3/day]
Desal plant capacity expansion
1096600 [m3/day]
Desal plant capacity expansion
931300 [m3/day]
Desal plant capacity expansion
914090 [m3/day]
Desal plant capacity expansion
273230 [m3/day]
Pipeline construction
Ras Al-Khair Riyadh
Pipeline construction
Medina Ha’il
Pipeline construction
Medina Ar-Rass
Pipeline construction
Tabuk Dumat Al-Jandal
Power plant construction
Ras Al-Khair
793 [MW]
Summary and future work
To summarize, we have developed a new multi-commodity network flow model, which is capable of
investigating individual resource flows at the facility level and of providing Pareto-optimal investment
portfolios, each of which comes with useful information in various forms such as resulting network flow and a
list of projects for investment purposes as illustrated above. Since the credibility of the result is driven by the
accuracy of input parameters and assumptions, it is important to collect more reliable data about the current
and future supply/demand, technology, and cost, part of which could be provided by single-facility analysis.
In order to consolidate this network model, a more realistic pipeline cost model is under development to
capture not only pipeline distance and diameter but also tunnel options. While the present model considers
pumping energy requirements due to friction and elevation difference, it assumes on-the-ground pipelines
that run along the topography of the surface. For example, the result presented above suggests a long-distance
pipeline between Medina and Ar-Rass. While the elevations of Medina and Ar-Rass are 613 m and 693 m,
respectively, there is a mountain region in between with the highest point up to 1130 m. Therefore, we must
consider either (1) laying a pipeline on the ground, or (2) drilling a tunnel for the whole or part of the distance.
If we drill a tunnel, the same assumption about pipeline CAPEX does not hold. Therefore, the model needs the
flexibility to allow for partially underground pipelines and the corresponding cost models.
The INFINIT model described up to this point discussed a static network flow problem, in which the flow
is optimized with respect to a given snapshot of supplies and demands. A static INFINIT model can solve an
optimization problem in spatial dimension, that is,
a new infrastructure should be invested at a given
time. However, it does not answer the question of
it should be. Since demand must be satisfied each year
(not only once) and a facility, once built at a certain time, operates during its lifetime of a few decades from
then, the problem to be solved is not a one-time optimization. The INFINIT model must be extended to the
temporal dimension so that it can optimize not only snapshots of future infrastructure but also optimal staged
deployment of future infrastructure projects over time. In other words, the transition of network topology
occurs in a staged manner, not at one time. This is one of the most important future work and once the
temporal dimension has been implemented, it will be able to determine where a new facility should be
established both in spatial and temporal dimensions. If the result shows that a new infrastructure is not
invested in the same site as an existing facility, then it implies that the site should be closed before or at the
end of its current lifetime. This would suggest an optimal strategy for staged transition of infrastructure
4. Visualization tools
In order to effectively visualize alternative design and policy scenarios, the DSS we propose in this study
allows two ways of visualization of the results: a MATLAB-based graphical user interface (GUI) and a tabletop
3D map projection. Fig. 9 shows a screenshot of the MATLAB-based GUI. On the left panel, the user can set
some computational parameters such as weighting between cost and CO2 emission, a computational time limit,
and a desired optimality gap. A resulting network graph is shown in the middle, and quantitative results are
shown in the right panel.
Some other choices using display visualization technologies are difficult to understand particularly
decisions relating to the topography of the region. In order to facilitate a better decision making, we also
created a topographic map of the Arabian Peninsula including KSA and the surrounding geographic areas of
neighboring countries which may be relevant to decisions to be made. As shown in Fig. 10, some areas further
from Saudi Arabia have been removed to create some flat areas around, onto which information such as a
legend can be projected.
Fig. 11 shows how the 3D map projection looks. The ultimate goal of this tool is an interactive system
where the users can manually place physical components representing new infrastructure elements or remove
existing ones and quickly simulate the optimal resource flows to see the impacts of this change on the whole
network at a national or regional level.
Fig. 9. MATLAB-based graphical user interface (GUI).
Fig. 10. Area covered in the KSA topographic map.
Fig. 11. KSA 3D map projection.
5. Conclusion
The water/energy network model views a single desalination/power plant as part of a larger network
representing a region or spanning the entire Kingdom of Saudi Arabia. Since desalination/power plants are
increasingly connected together with water pipelines and electricity transmission lines, evaluating a stand-
alone plant as a node in a larger network is compelling. The water/energy network analysis can investigate
and identify candidate locations for sustainable desalination infrastructure investments, accounting for the
existing assets and the current investment plans.
We presented in this paper a network-based DSS, allowing the optimization and visualization of
alternative design and policy scenarios addressing the location of water and energy investments across the
Kingdom. For the underlying mathematical model, a new multi-commodity network flow modeling method
called INFINIT was employed. The INFINIT network model allows for optimizing individual resource flows
at the facility level and providing Pareto-optimal investment portfolios. As a use case, a static network flow
optimization was performed for 2030 demand with 2010 infrastructure, and the graphical and quantitative
results were presented. In addition to a MATLAB-based GUI, we also introduced a tabletop 3D map projection
for visualizing plants and network assets across the Kingdom with topographical consideration.
The authors wish to acknowledge the support of Anas Alfaris, Adnan Alsaati, Kenneth Strzepek, and
Stephen Connors. The authors also wish to thank Abdulaziz Alhassan, Abdulaziz Khiyami, David Barmore,
Malak Al-Nory, Salma Aldawood, Vivek Sakhrani, and all researchers at KACST and MIT (http://www.cces- for having shared their data, expectations, and feedback.
[1] H.M.H. Al-Sheikh, Appendix 8: Country case study water policy reform in Saudi Arabia, Proceedings of
the second expert consultation on national water policy reform in the Near East (1997).
[2] MoWE, The Saudi Arabian Ministry of Water and Electricity annual technical report (2013).
[3] E. Kondili, J.K. Kaldellis, C. Papapostolou, A novel systemic approach to water resources optimisation in
areas with limited water resources, Desalination 250 (2010) 297-301.
[4] M.T. Al-Nory, S.C. Graves, Water desalination supply chain modeling and optimization: case of Saudi
Arabia, IDA Journal of Desalination and Water Reuse 5(2) (2013) 64-74.
[5] M.T. Al-Nory, A. Brodsky, B. Bozkaya, S.C. Graves, Desalination supply chain decision analysis and
optimization, Desalination 347 (2014) 144-157.
[6] T. Ishimatsu, Generalized Multi-Commodity Network Flows: Case Studies in Space Logistics and Complex
Infrastructure Systems, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA,
Aeronautics and Astronautics. Doctor of Philosophy thesis, 2013.
[7] T. Ishimatsu, O.L. de Weck, Interdependent Multicommodity Network Flow Modeling Framework,
Operations Research (2017). (Under review)
[8] H.G. Chale-Gongora, O. de Weck, A. Doufene, T. Ishimatsu, D. Krob, Planning an Itinerary for an Electric
Vehicle, Proceedings of the IEEE International Energy Conference (EnergyCon) (2014).
[9] T. Ishimatsu, O.L. de Weck, J.A. Hoffman, Y. Ohkami, R. Shishko, Generalized Multicommodity Network
Flow Model for the Earth-Moon-Mars Logistics System, Journal of Spacecraft and Rockets 53(1) (2016)
Table 3
Input parameters and assumptions used in this analysis.
Assumed value
Ratio of non-potable water demand to water demand
Ratio of wastewater generated to potable water consumed
Ratio of non-potable water generated to wastewater generated
Power consumption for water desalination by technology [kWh/m3]
Sea water reverse osmosis (SWRO)
Brackish water reverse osmosis (BWRO)
Multi-stage flash distillation (MSF)
Multiple-effect distillation (MED)
Power consumption for groundwater processing [kWh/m3]
Power consumption for wastewater treatment [kWh/m3]
Friction head loss coefficient of water transmission
Typical water transmission speed [m/s]
Water loss during transmission [%/km]
Resistive loss of electricity transmission [%/km]
CAPEX (annualized) for:
Desalination plant capacity expansion [USD/(m3/day)/year]
Power plant capacity expansion [USD/MW/year]
Groundwater capacity expansion [USD/(m3/day)/year]
Wastewater treatment capacity expansion [USD/(m3/day)/year]
Water pipeline [USD/km/year]
Powerline [USD/km/year]
OPEX for:
Desalination plant [USD/m3]
Power plant [USD/kWh]
Groundwater processing [USD/m#]
Wastewater treatment [USD/m3]
Water pipeline [USD/km/year]
Powerline [USD/km/year]
CO2 emission by:
Water desalination by technology [kg/m3]
Sea water reverse osmosis (SWRO)
Brackish water reverse osmosis (BWRO)
Multi-stage flash distillation (MSF)
Multiple-effect distillation (MED)
Groundwater processing [kg/m3]
Wastewater treatment [kg/m3]
Power generation by fuel [kg/kWh]
Fig. 12. List of major cities used in this analysis.
Potable Non-Potable
Haql Tabuk 1 29.286097 34.938583 47 25649 39697 5530 2370 6320
Tabuk Tabuk 2 28.390409 36.573200 769 512629 793398 110520 47366 126309
Tayma Tabuk 3 27.627741 38.560120 827 30411 47067 6556 2810 7493
Duba Tabuk 4 27.349260 35.696190 6 25568 39572 5512 2362 6300
Al-Wajh Tabuk 5 26.228790 36.469059 2 30512 47224 6578 2819 7518
Umluj Tabuk 6 25.050006 37.265108 23 37757 58437 8140 3489 9303
Al-Ula Madinah 7 26.612829 37 .922918 778 32413 50166 8147 3492 9311
Khaybar M adinah 8 25.698611 39.292500 774 45489 70403 11434 4900 13067
Medina Madinah 9 24.460900 39.620190 613 1100093 1702618 276505 118502 316006
Yanbu Madinah 10 24.086700 38.058552 11 233236 360980 58623 25124 66998
Badr Madinah 11 23.780000 38.790556 130 28999 44882 7289 3124 8330
Masturah Makkah 12 23.110901 38.850689 12 266 412 62 26 70
Rabigh Makkah 13 22.800000 39.033333 6 55304 85594 12822 5495 14654
Thuwal Makkah 14 22.283333 39.100000 4 274 424 64 27 73
Al-Jumum Makkah 15 21.619890 39.699587 229 25601 39623 5935 2544 6783
Jeddah Makkah 16 21.543489 39.172989 15 3430697 5309702 795393 340883 909021
Al-Hawiyah Makkah 17 21.441111 40.497500 1508 148151 229294 34348 14721 39255
Mecca Makkah 18 21.416667 39.816667 330 1534731 2375309 355821 152495 406653
Bahrah Makkah 19 21.401667 39.450833 97 75213 116407 17438 7473 19929
Al-Khurmah Makkah 20 21.273232 40.427258 1669 27032 41838 6267 2686 7163
Taif Makkah 21 21.270660 40.417068 1676 579970 897622 134464 57627 153673
Ranyah Makkah 22 21.262738 42.854040 920 21656 33517 5021 2152 5738
Turabah Makkah 23 21.222639 41.647301 1157 25937 40143 6013 2577 6872
Al-Qunfudhah Makkah 24 19.128142 41.078739 3 24512 37937 5683 2436 6495
Al-Qouz Makkah 25 18.980531 41.305950 47 23391 36202 5423 2324 6198
Hali Makkah 26 18.700000 41.333333 22 1000 1548 232 99 265
Al-Bahah Al-Bahah 27 20.012991 41.460449 2175 95089 147170 6181 2649 7064
Baljurashi Al-Bahah 28 19.858062 41.571230 2031 43493 67314 2827 1212 3231
Al-Makhwah Al-Bahah 29 19.756040 41.436246 378 21999 34048 1430 613 1634
Bisha Asir 30 20.000000 42.600000 1165 86201 133414 7751 3322 8859
Sabt Alalayah Asir 31 19.597836 41.978993 1975 19956 30886 1794 769 2051
An-Nimas Asir 32 19.098994 42.127934 2420 27021 41821 2430 1041 2777
Muhayil Asir 33 18.547395 42.053440 520 56953 88146 5121 2195 5853
Khamis Mushayt Asir 34 18.300000 42.733333 2004 430828 666794 38741 16603 44275
Abha Asir 35 18.220221 42.508160 2215 236157 365501 21236 9101 24269
Ahad Rafidah Asir 36 18.212171 42.844291 2079 57112 88392 5136 2201 5869
Dhahran Al-Janub Asir 37 17.665225 43.517421 2156 23758 36770 2136 916 2442
Baish Jazan 38 17.374003 42.536250 78 30835 47723 1169 501 1336
Sabya Jazan 39 17.148992 42.625923 36 63143 97727 2394 1026 2736
Damad Jazan 40 17.106380 42.777481 80 24056 37232 912 391 1042
Abu Arish Jazan 41 16.968889 42.832500 67 61047 94483 2315 992 2646
Jazan Jazan 42 16.891920 42.549751 6 127743 197708 4844 2076 5536
Ahad Al-Masarihah Jazan 43 16.709722 42.955000 80 25007 38703 948 406 1084
Samtah Jazan 44 16.597222 42.943889 62 32458 50235 1231 527 1407
Najran Najran 45 17.491730 44.132290 1313 298288 461661 17451 7479 19944
Sharurah Najran 46 17.483333 47.116667 727 75237 116445 4402 1886 5030
Hafr Al-Batin Eastern Province 47 28.434151 45.975300 315 271642 420421 101237 43387 115700
Khafji Eastern Province 48 28.416667 48.500000 1 67012 103715 24975 10703 28542
Al-Qaisumah Eastern Province 49 28.309722 46.127500 363 22538 34882 8400 3600 9600
Nairyah Eastern Province 50 27.472340 48.481171 51 26470 40968 9865 4228 11274
Jubail Eastern Province 51 27.012563 49.658128 0 337778 522780 125885 53951 143869
Ras Tanura Eastern Province 52 26.701864 50.049572 3 54166 83833 20187 8652 23071
Safwa Eastern Province 53 26.650000 49.950000 14 50447 78077 18801 8058 21487
Tarout Eastern Province 54 26.566667 50.066667 5 77757 120345 28979 12420 33119
Qatif Eastern Province 55 26.559050 49.995689 7 172922 267632 64446 27620 73652
Ank Eastern Province 56 26.519577 50.026727 6 23125 35791 8618 3694 9850
Saihat Eastern Province 57 26.475000 50.041667 6 75794 117307 28247 12106 32283
Dammam Eastern Province 58 26.392666 49.977714 5 903312 1398059 336653 144280 384746
Dhahran Eastern Province 59 26.288768 50.114103 46 120521 186531 44917 19250 51333
Khobar Eastern Province 60 26.283333 50.200000 16 457745 708454 170596 73112 194966
Abqaiq Eastern Province 61 25.933335 49.666667 99 36207 56038 13494 5783 15422
Al-Uyun Eastern Province 62 25.600000 49.566667 113 33042 51139 12314 5278 14074
Hofuf Eastern Province 63 25.383333 49.583333 153 660788 1022704 246267 105543 281448
Al-Taraf Eastern Province 64 25.352520 49.724330 137 23543 36438 8774 3760 10028
Al-Zilfi Riyadh 65 26.313754 44.849229 700 60867 94204 19255 8252 22006
Al-Ghat Riyadh 66 26.026668 44.960833 706 8497 13151 2688 1152 3072
Al-Majma'ah Riyadh 67 25.910530 45.358860 712 47743 73892 15104 6473 17261
Sudair Riyadh 68 25.601243 45.630314 730 14316 22157 4529 1941 5176
Rumah Riyadh 69 25.563706 47.160584 565 20276 31381 6414 2749 7331
Shaqra Riyadh 70 25.249359 45.261669 712 26651 41248 8431 3613 9635
Ad-Diriyah Riyadh 71 24.733222 46.564421 654 43269 66968 13688 5866 15644
Riyadh Riyadh 72 24.711667 46.724167 635 5188286 8029929 1641318 703422 1875792
Dawadmi Riyadh 73 24.504444 44.393889 977 61834 95701 19561 8383 22356
Muzahmiyya Riyadh 74 24.481470 46.255760 629 30164 46685 9542 4090 10906
Al-Kharj Riyadh 75 24.148330 47.305000 438 234607 363102 74218 31808 84821
Al-Quwaiiyah Riyadh 76 24.046339 45.265612 854 22519 34853 7124 3053 8142
Dilam Riyadh 77 23.993521 47.174129 446 40114 62085 12690 5439 14503
Afif Riyadh 78 23.910000 42.920278 1048 45525 70459 14402 6172 16459
Hawtah Bani Tamim Riyadh 79 23.525188 46.844661 616 26270 40658 8311 3562 9498
Layla Riyadh 80 22.283333 46.733333 538 30906 47833 9777 4190 11174
Wadi Ad-Dawasir Riyadh 81 20.471998 44.785584 684 93036 143992 29432 12614 33637
As-Sullayyil Riyadh 82 20.459722 45.574444 606 26639 41229 8427 3612 9631
Uyun Al-Jiwa Al-Qassim 83 26.526068 43.687890 660 33042 51139 9415 4035 10760
Buraydah Al-Qassim 84 26.333333 43.966667 624 467410 723412 133180 57077 152206
Al-Bukairyah Al-Qassim 85 26.126419 43.690880 658 29547 45730 8419 3608 9622
Unaizah Al-Qassim 86 26.084761 43.994301 665 152895 236636 43565 18671 49788
Albadaye Al-Qassim 87 25.981180 43.733409 668 46620 72154 13284 5693 15181
Ar-Rass Al-Qassim 88 25.868340 43.512890 693 92501 143164 26357 11296 30122
Al-Mithnab Al-Qassim 89 25.852668 44.230216 632 29210 45208 8323 3567 9512
Ha'il Ha'il 90 27.516667 41.683333 1007 310897 481176 45808 19632 52352
Turaif Northern Borders 91 31.674191 38.656399 828 48108 74457 7453 3194 8518
Ar'ar Northern Borders 92 30.980829 41.025009 538 167057 258555 25881 11092 29579
Rafha Northern Borders 93 29.633301 43.510929 447 52712 81583 8166 3500 9333
Al-Qurayyat Al-Jouf 94 31.312979 37.374888 501 116162 179784 30078 12891 34375
Tabarjal Al-Jouf 95 30.500000 38.216667 546 48525 75102 12565 5385 14360
Sakakah Al-Jouf 96 29.970869 40.205009 558 150257 232553 38906 16674 44464
Dumat Al-Jandal Al-Jouf 97 29.818202 39.868198 600 32613 50475 8445 3619 9651
[deg N]
[deg E]
Water Demand [m3/day]
Fig. 13. List of major desalination/power plants used in this analysis.
Haql D SWCC Tabuk 101 29.290733 34.929889 11
sea water SWRO 6000 0
Duba D SWCC Tabuk 102 27.362429 35.662570 11
brackish water BWRO 42250 0
Al-Wajh D SWCC Tabuk 103 26.239238 36.450558 6
sea water MED 18100 0
Al-Wajh 4 D SWCC Tabuk 151 26.239238 36.450558 6
sea water MED 0 0
Umluj MED D SWCC Tabuk 104 25.002671 37.274977 4
sea water MED 9000 0
Umluj RO D SWCC Tabuk 105 25.002671 37.274977 4
river water BWRO 35000 0
Yanbu 2 DP SWCC Madinah 106 23.872239 38.364825 6
sea water M SF 144000 HFO 162.8
Yanbu RO D SWCC Madinah 107 23.872239 38.364825 6
sea water SWRO 127800 0
Yanbu 1 DP SWCC Madinah 108 23.867581 38.369019 1
sea water M SF 0 HFO 0
Yanbu 3 DP SWCC Madinah 152 23.872239 38.364825 6
sea water M SF 0 HFO 0
Yanbu MED D SWCC Madinah 153 23.872239 38.364825 6
sea water MED 0 0
Yanbu 1 D Marafiq Madinah 154 23.972716 38.215687 8
sea water MED 0 0
Yanbu 2 DP Marafiq Madinah 155 23.972716 38.215687 8
sea water MED 0 HFO 0
Rabigh D SWCC Makkah 109 22.778950 38.961911 0
sea water MED 18000 0
Rabigh D Huta Makkah 110 22.778950 38.961911 0
sea water SWRO 30000 0
Rabigh 4 D SWCC Makkah 156 22.77895 38.961911 0
sea water SWRO 0 0
Aziziyah D SWCC Makkah 111 22.095696 39.032558 6
sea water MED 4500 0
Jeddah 4 DP SWCC M akkah 112 21.553061 39.116313 11
sea water M SF 0 HFO 0
Jeddah 3 DP SWCC M akkah 113 21.550639 39.114814 10
sea water M SF 0 HFO 0
Jeddah RO2 D SWCC Makkah 114 21.553061 39.116313 11
sea water SWRO 56800 0
Jeddah RO1 D SWCC Makkah 115 21.553061 39.116313 11
sea water SWRO 56800 0
Jeddah RO D Makkah 116 21.553061 39.116313 11
sea water SWRO 27600 0
Jeddah WW D GOV Makkah 117 21.543489 39.172989 15
waste water SWRO 30000 0
Jeddah RO3 D SWCC Makkah 157 21.553061 39.116313 11
sea water SWRO 0 0
Shoaiba 3 DP SWEC Makkah 118 20.681198 39.520305 7
sea water M SF 880000 HFO 1190.7
Shoaiba Ex D SEPC Makkah 119 20.676091 39.523210 8
sea water SWRO 150000 0
Shoaiba 2 DP SWCC Makkah 120 20.671396 39.526242 9
sea water M SF 454000 HFO 520
Shoaiba 1 DP SWCC Makkah 121 20.627626 39.555765 7
sea water M SF 223000 HFO 262.8
Shoaiba Barge D RAKA Makkah 122 20.627626 39.555765 7
sea water SWRO 52000 0
Al-Lith D SWCC Makkah 123 20.150511 40.266850 3
sea water MED 9000 0
Qunfudhah D SWCC Makkah 124 19.075763 41.164482 5
sea water MED 9000 0
Al-Birk D SWCC Makkah 125 18.209489 41.525435 5
sea water SWRO 2200 0
Shuqaiq 2 DP SqWEC Jazan 126 17.658759 42.076809 7
sea water SWRO 213000 crude 1020
Shuqaiq 1 DP SWCC Jazan 127 17.660904 42.063008 14
sea water M SF 97014 HFO 128
Khafji D SWCC Eastern Province 128 28.414542 48.530718 6
sea water M SF 36000 0
Ras Al-Khair DP SWCC Eastern Province 158 27.539057 49.196598 2
sea water M SF 0 gas 0
Ras Al-Khair RO D SWCC Eastern Province 159 27.539057 49.196598 2
sea water SWRO 0 0
Jubail MSF DP SWCC Eastern Province 129 26.897219 49.778863 10
sea water M SF 0 HFO 0
Jubail RO D SWCC Eastern Province 130 26.897219 49.778863 10
sea water SWRO 261809 0
Jubail RO2 D Marafiq Eastern Province 160 26.897219 49.778863 10
sea water SWRO 0 g as 0
Khobar 2 DP SWCC Eastern Province 131 26.178118 50.210179 12
sea water M SF 0 gas 0
Khobar 3 DP SWCC Eastern Province 132 26.175467 50.207548 9
sea water M SF 280000 HFO 478.8
Energy Source
Water Capacity
Power Capacity
Desal/Power Plant
Feed Water
[deg N]
[deg E]
Fig. 14. List of major power plants used in this analysis.
Tabuk 1 P SEC Tabuk 201 28.330835 36.584622 767 Diesel 134.20
Tabuk 2 P SEC Tabuk 202 28.469697 36.517380 751 Diesel 335.60
Duba P SEC Tabuk 203 27.367544 35.659148 12 Diesel 162.00
Al-Wajh P SEC Tabuk 204 26.223823 36.488845 8 Diesel 86.00
Medina P SEC Madinah 205 24.460900 39.620190 613 Diesel 452.80
Yanbu P Aramco Madinah 206 23.970644 38.262468 7 Gas 82.50
Yanbu P Marafiq G Madinah 207 23.991673 38.232960 8 Gas 524.60
Yanbu P Marafiq H Madinah 208 23.991673 38.232960 8 HFO 508.00
Yanbu P SEC Madinah 209 24.086700 38.058552 11 Diesel 54.50
Rabigh P SEC C Makkah 210 22.800000 39.033333 7 Crude 3113.46
Rabigh P SEC H Makkah 211 22.800000 39.033333 7 HFO 1575.60
Jeddah P SEC D Ma kkah 212 21.543489 39.172989 15 Diesel 806.90
Jeddah P SEC C Makkah 213 21.543489 39.172989 15 Crude 1630.00
Mecca P SEC Makkah 214 21.416670 39.816670 330 Diesel 1147.10
Taif P SEC Makkah 215 21.282224 40.406195 1673 Diesel 200.40
Shoaiba P SEC Makkah 216 20.627516 39.555705 7 HFO 4323.00
Tihama P SEC Makkah 217 19.095824 41.157887 10 Diesel 697.09
Al-Bahah P SEC Al-Bahah 218 20.012991 41.460449 2175 Diesel 85.60
Bisha P SEC Asir 219 19.987241 42.393704 1220 Diesel 365.40
Asir CPS P SEC Asir 220 18.246183 42.578331 2172 Diesel 639.72
Baish P SEC Jazan 221 17.374003 42.536250 78 Diesel 26.40
Jazan P SEC Jazan 222 16.937474 42.633229 24 Diesel 1353.65
Samtah P SEC Jazan 223 16.597222 42.943889 62 Diesel 25.00
Najran P SEC Najran 224 17.596377 44.337370 1254 Diesel 436.00
Sharurah P SEC Najran 225 17.331103 47.093045 759 Diesel 105.92
Qaisumah P SEC Eastern Province 226 28.348841 46.048965 369 Diesel 153.80
Safaniyah P SEC Eastern Province 227 28.000165 48.753340 10 Gas 94.80
Khursaniyah P Aramco Eastern Province 228 27.065248 49.261289 24 Gas 298.00
Jubail P JEC Eastern Province 229 27.055052 49.596196 14 Gas 250.00
Jubail P JWPC Eastern Province 230 27.055052 49.596196 14 Gas 733.33
Berri P Aramco Eastern Province 231 26.956166 49.589445 18 Gas 298.00
Berri P SEC Eastern Province 232 26.956166 49.589445 18 Gas 278.10
Ghazlan P SEC Eastern Province 233 26.854569 49.895607 13 Gas 4256.00
Juaymah P SEC G Eastern Province 234 26.793366 50.006868 7 Gas 158.70
Juaymah P SEC D Eastern Province 235 26.793366 50.006868 7 Diesel 10.80
Juaymah P TPGC Eastern Province 236 26.793366 50.006868 7 Gas 310.00
Qatif P Aramco Eastern Province 237 26.788265 49.944084 5 Gas 144.00
Ras Tanura P TPGC Eastern Province 238 26.698637 50.097106 6 Gas 153.00
Dammam P SEC Eastern Province 239 26.381623 50.093836 47 Gas 582.50
Abqaiq P Aramco Eastern Province 240 25.928889 49.687778 97 Gas 129.00
Ain Dar P SCC Eastern Province 241 25.925676 49.469298 146 Gas 76.80
Qurayyah P SEC Eastern Province 242 25.859581 50.115573 6 Gas 4532.00
Shedgum P SEC Eastern Province 243 25.652018 49.391654 298 Gas 1429.50
Shedgum P TPGC Eastern Province 244 25.652018 49.391654 298 Gas 310.00
Hofuf P SCC G Eastern Province 245 25.332825 49.722984 138 Gas 81.00
Hofuf P SCC C Eastern Province 246 25.332825 49.722984 138 Crude 108.00
Uthmaniyah P SEC Eastern Province 247 25.312005 49.343107 293 Gas 41 2.20
Uthmaniyah P TPGC Eastern Province 248 25.312005 49.343107 293 Gas 310.00
Faras P SEC Eastern Province 249 25.210032 49.285314 261 Gas 1568.70
Riyadh PP9 P SEC Riyadh 250 24.945655 47.065096 668 Gas 3760.60
Riyadh PP5 P SEC Riyadh 251 24.759032 46.592737 680 Crude 608.00
Riyadh PP3 P SEC Riyadh 252 24.652952 46.726875 590 Diesel 65.00
Riyadh PP4 P SEC Riyadh 253 24.652540 46.672353 610 Diesel 336.49
Riyadh PP8 P SEC Riyadh 254 24.597212 46.571972 740 Gas 2060.50
Riyadh PP7 P SEC Riyadh 255 24.569941 46.885403 566 Gas 1315.78
Riyadh P Aramco Riyadh 256 24.522474 46.866235 567 Gas 66.00
Riyadh PP10 P SEC Riyadh 257 24.415381 47.013971 511 Crude 1118.00
Layla P SEC Riyadh 258 22.309044 46.662492 558 Crude 102.00
Juba P SEC C Riyadh 259 20.392970 45.206759 630 Crude 230.35
Juba P SEC D Riyadh 260 20.392970 45.206759 630 Diesel 100.08
Buraydah P SEC Al-Qassim 261 26.402126 43.944573 624 Diesel 104.50
Qassim Central P SEC Al-Qassim 262 26.203288 44.014917 622 Crude 1138.06
Ha'il 1 P SEC Ha'il 263 27.535063 41.701593 994 Diesel 48.40
Ha'il 2 P SEC Ha'il 264 27.468809 41.741723 1004 Crude 340.40
Ar'ar P SEC Northern Borders 265 30.927203 41.055819 545 Diesel 273.80
Rafha P SEC Northern Borders 266 29.619087 43.526290 458 Diesel 126.40
Qurayyat P SEC Al-Jouf 267 31.251532 37.427176 517 Diesel 116.80
Tabarjal P SEC Al-Jouf 268 30.455557 38.212876 552 Diesel 99.90
Al-Jouf P SEC Al-Jouf 269 29.777394 40.011580 673 Crude 238.00
Power Plant
[deg N]
[deg E]
Energy Source
Feed Water
Water Capacity
Power Capacity
... Similarly, some works have addressed the dependence of the electric grid on the oil [51], [66] and coal systems [52], [67]. Moving beyond the specific scope of the AMES, a related but extensive literature has developed on the co-dependence of the electric grid and water resources in the form of the Energy Water Nexus (EWN) [15], [68]- [81]. Together, these works provide an insight into the structural and behavioral complexity of the AMES. ...
Full-text available
The American Multimodal Energy System (AMES)is a systems-of-systems comprised of four separate but interdependent infrastructure systems: the electric grid, the natural gas system, the oil system, and the coal system. Their interdependence creates the need to better understand the underlying architecture in order to pursue a more sustainable, resilient and accessible energy system. Collectively, these requirements necessitate a sustainable energy transition that constitute a change in the AMES' instantiated architecture; although it leaves its reference architecture largely unchanged. Consequently, from a model-based systems engineering perspective, identifying the underlying reference architecture becomes a high priority. This paper defines a reference architecture for the AMES and its four component energy infrastructures in a single SysML model. The architecture includes (allocated) block definition and activity diagrams for each infrastructure. The reference architecture was developed from the S&P Global Platts (GIS) map data pro data set and the EIA Annual Energy Outlook dataset.
... So, in summary, DSS is mostly used for effective and optimal planning and allocation of resources in membrane-based plants. In some cases, it was used for optimal selection of the location to erect a desalination plant and the type of water treatment technology that would ensure a sustainable operation (Abdulbaki et al., 2017;Aliewi et al., 2017;Geza et al., 2018;Ishimatsu et al., 2017;. DSS has also been used to select the technology with the minimum environmental footprint. ...
Access to clean and potable water will continue to be a global challenge in as much as sustainable solutions are far-fetched. It has become highly imperative to improve the efficiency of conventional membrane science and technologies for water treatment in order to reduce their deleterious impact on the environment. Some sustainability solutions have been proposed and studied in the past decade. Therefore, a critical and comprehensive review of emerging trends in sustainable membrane-based desalination and wastewater treatment is presented in this paper. Some of the emerging trends in membrane science and technology for sustainable desalination and circular economy solutions include the reuse of membranes, reuse of waste brine or sludge, energy harvesting from wastes, and waste reduction by membrane antifouling approaches. RO membranes that have reached their end-of-life are reused as UF and NF membranes whereas extremely damaged membranes are used in membrane biofilm reactors or as support materials for recycled anion-exchange membranes. There is more research on processes integrated with PRO for energy harvesting from wastes. The use of membrane-based ZLD approaches are also being intensified for enhanced water recovery, solute recovery, and recovery of precious metals and chemicals from wastewater and desalination concentrates. More sustainable materials such as QDs and green solvents for membrane synthesis are being developed but more research with respect to their toxicity should be carried out. Many traditional membrane synthesis methods such as casting, coating, grafting, and vapor deposition do not support rapid prototyping. 3D and 4D printing has attracted recent research attention for rapid prototyping and flexibility in the manufacturing of membrane module materials. AI tools have also been presented in recent studies as effective future solutions for decision making and for the prediction, operation, and control of membrane-based water treatment processes. The challenges associated with these emerging trends are also discussed in this review.
... Water and energy consumption cannot be disconnected, since they display a direct dependence [86]. From this perspective, the analysis of these links should be able to: (i) identify ideal locations for desalination use; (ii) consider infrastructure investments, accounting for existing assets and (iii) assess the best allocation of water and energy investments [87]. ...
Full-text available
Semi-arid regions have historically suffered from low water availability. In addition, the increasing frequency and intensity of extreme weather events credited to global climate change has made it increasingly clear that among the challenges faced by society water resource management is extremely necessary. In this context, desalination based on renewable energy resources integrated with production systems that make use of the waste resulting from this process becomes a socio-environmentally indicated alternative to expand existing supply strategies and sustainable water use in isolated locations, and/or areas distant from large urban centers, thus addressing local potential and reducing environmental impacts. This study assesses the use of Photovoltaic Solar Power Plants (PSPPs), as well as of residues generated in a Brackish Water Reverse Osmosis System (BWRO), in productive units linked to fish and family farming. This is as an alternative way to reduce water vulnerability in the Brazilian semi-arid area (BS), adhering to climate change adaptation measures in the light of Brazilian public policies through the Freshwater Program (Programa Água Doce—PAD), which aims to promote access to good quality water to approximately 500 thousand people in the Brazilian semi-arid region.
... conducted a single facility [21] analysis to nd the best location of a desalination plant driven by the fi renewable energy as its primary source. Considering the geographical aspect of the problem, they used a MILP-based model to minimise both total cost and CO 2 emissions to nd a solution at national scale (Saudi fi Arabia) in 2030. ...
An interactive multi-period planning model is presented for sustainable urban water and energy supply, taking into account surplus output from grid-connected residential photovoltaics as a part of the water-related energy mix. The two-level mixed integer linear model finds the optimal strategic and operational decisions for a desalination-based water supply system driven by hybrid energy sources and determines the evolution of the potential capacity of a renewable energy technology over the planning horizon. It considers demands, supply systems configuration, resources capacities and electricity tariffs as well as economic, subjectivity and technical criteria for uptaking rooftop photovoltaic systems. The model was then applied to Perth (Australia) and solved for alternative scenarios. The results show operational flexibility and decentralised planning of the integrated system lead to $251,515,132 less discounted total cost over centralised water supply system operated in fixed mode. They also indicate that decentralised scenario results in 42,765.1 kW higher potential photovoltaics uptake capacity on average in each year over the planning horizon in the case study area compared to centralised scenarios. However, based on the results of the sensitivity analysis, the selection of this scenario as the best alternative highly depends on the parameters values associated with subjectivity criterion and operational and maintenance cost of flexible mode of operation.
Sustainable energy grids and networks are converging. In recent decades, cost, sustainability, and digitization drivers have caused engineering systems to evolve into systems-of-systems that deliver multiple services across multiple application domains. These engineering systems include electrified-transportation systems, the energy–water nexus, and multi-modal energy systems. The rather complex and heterogeneous interdependencies between engineering system services necessitates a precise informatic representation that ultimately supports optimized management of the holistic dynamics and tradeoffs. Consequently, ontologically-robust, quantitative modeling tools are needed to represent the heterogeneity of the modeled system-of-systems while still remaining generic and extensible to a diversity of application domains. Hetero-functional graph theory has demonstrated itself as a such modeling tool. This work now builds upon this foundation to develop a dynamic hetero-functional network minimum cost flow optimization that meets the requirements of these emerging systems-of-systems. It optimizes the supply, demand, transportation, storage, transformation, assembly, and disassembly of multiple operands in distinct locations over time in a systems-of-systems of arbitrary number, function, and topology. First, the paper introduces a general approach to define a dynamic system-of-system model that integrates customizable dynamic device models into a hetero-functional graph theory structural model. To this end, the work leverages Petri net dynamics and the hetero-functional incidence tensor. The Petri net based models are then translated into a quadratic program in canonical form. The hetero-functional network minimum cost flow optimization is demonstrated on a hydrogen–natural gas infrastructure test case. Four distinct scenarios are studied to demonstrate potential synergies and cascading network effects of policy across multiple infrastructures.
Artificial intelligence, an emerging technology, widely exists in the field of engineering science and technology. Due to its high efficiency and precision, artificial intelligence is increasingly used in the optimal control of water treatment and seawater desalination. Generally, the design of a desalination system includes four processes: site selection, energy prediction, desalination technology selection and systematic parameter optimization. To a large extent, these choices depend on the experience and relevant criteria of researchers and experts. However, facing the scientific and technological progress and growing expectations, it is impossible to solve such complex nonlinear problems by simple experience and mathematical models, but artificial intelligence is good at this. In this paper, we synthetically analyzed and summarized the application of artificial intelligence in the field of seawater desalination with renewable energy. Artificial intelligence application in desalination is mainly divided into four aspects: expert decision-making, optimization, prediction and control by sequence. The features of artificial intelligence employed in the design of desalination systems not only realize the maximum of efficiency and minimum of cost, but release the human resources. After analyzing the four processes of desalination, it is found that artificial neural network and genetic algorithm are more widespread and mature than other algorithms in dealing with multi-objective nonlinear problems. This paper overviewed the application of artificial intelligence technologies in decision-making, optimization, prediction and control throughout the four processes of desalination designs. Finally, the application and future development prospect of artificial intelligence in the field of seawater desalination are summarized.
Full-text available
Over the past decades, engineering systems have developed as networks of systems that deliver multiple services across multiple domains. This work aims to develop an optimization program for a dynamic, hetero-functional graph theory-based model of an engineering system. The manuscript first introduces a general approach to define a dynamic system model by integrating the device models in the hetero-functional graph theory structural model. To this end, the work leverages Petri net dynamics and the hetero-functional incidence tensor. The respective Petri net-based models are translated into the quadratic program canonical form to finalize the optimization program. The optimization program is demonstrated through the application of the program to a hydrogen-natural gas infrastructure test case. Four distinct scenarios are optimized to demonstrate potential synergies or cascading network effects of policy across infrastructures. This work develops the first hetero-functional graph theory-based optimization program and demonstrates that the program can be used to optimize flows across a multi-operand network, transform the operands in the network, store operands over time, analyze the behavior for a quadratic cost function, and implement it for a generic, continuous, large flexible engineering systems of arbitrary topology.
This paper proposes a simulation model to evaluate an effectiveness of GHG reduction on ship transportation. Our model is extended INFINIT model, which is generalized multi-commodity network model and applicable for ship transportation system. The model makes it possible to evaluate the system with GHG emission, operating cost, and opportunity loss if the system’s capacity is short. Moreover, the method shows the detailed flow of ships, cargos, and fuels on the transportation network. As a case study, the method was applied for international transportation of iron ore. The case study demonstrates that the method can support decision-making by comparison of multiple scenarios to reduce GHG emission in shipping. Especially, it is useful in that it can evaluate those scenarios from the perspectives of not only transportation performance but also required bunkering infrastructures.
Water desalination combined with renewable energy sources (RES) constitutes an environmentally friendly technology for alleviating the scarcity of potable water. In this article, a new methodology is presented for calculating the optimal structures of desalination plants that are power-supplied by RES and cooperate with smart grids. The proposed technique takes into account the tradeoff between the three alternative degrees of freedom in the operation of the overall desalination plant, i.e., battery storage, water storage, and dynamic exchange of energy with the smart grid. The design results verify that the economically optimized configurations derived by applying the proposed design tool are capable to cover the water requirements of the consumers and support the operation of the electric grid by injecting the RES-generated energy surplus. Numerical results are also presented, demonstrating that the application of the proposed methodology enables to reduce the lifetime cost of the desalination plant by 60% compared with the exclusive use of electric grid energy. Also, by employing the total lifetime cost of the grid-connected RES-based desalination plant as objective function of the optimal design problem, the economical viability of the desalination system is improved.
Full-text available
Simple logistics strategies such as "carry-along" and Earth-based "resupply" were sufficient for past human space programs. Next-generation space logistics paradigms are expected to be more complex, involving multiple exploration destinations and in-situ resource utilization (ISRU). Optional ISRU brings additional complexity to the interplanetary supply chain network design problem. This paper presents an interdependent network flow modeling method for determining optimal logistics strategies for space exploration and its application to the human exploration of Mars. It is found that a strategy utilizing lunar resources in the cislunar network may improve overall launch mass to low Earth orbit for recurring missions to Mars compared to NASA’s Mars Design Reference Architecture 5.0, even when including the mass of the ISRU infrastructures that need to be pre-deployed. Other findings suggest that chemical propulsion using LOX/LH[subscript 2], lunar ISRU water production, and the use of aerocapture significantly contribute to reducing launch mass from Earth. A sensitivity analysis of ISRU reveals that under the given assumptions, local lunar resources become attractive at productivity levels above 1.8 kg/year/kg in the context of future human exploration of Mars.
Conference Paper
Full-text available
A large number of new desalination plants are being contracted every year constituting huge strategic investments. Strategic decisions related to plant locations and capacity, the selection of the desalination technology, and many other technical decisions related to the plant design and operation are very critical to maximize the economical and the social return on these investments. Viewing the desalination industry network as a supply chain provides a holistic view allowing decision makers to perform optimization of water desalination operations end to end. The proposed methodology provides a set of modular simulation components to allow the creation of complex models to optimize the entire water desalination supply chain quickly and easily. The optimization entails mathematical programming (MP) models that can be solved by external MP solvers. Saudi Arabia is the worldwide leader in the desalinated water capacity. A national water strategy is implemented by a number of governmental and privatized authorities engaged in major desalination and power generation projects aiming at covering 80% of municipal water demand by desalination by the end of 2014. We use the case of Saudi Arabia desalination supply chain to show the advantages of performing optimization using our proposed methodology. We solve a base case comprising the datasets collected from Saudi Arabia desalination authorities’ actual operational reports and the national development plans. We assume that the decision maker in this case wishes to find the optimal network flow given the specified demand. Then we show how the decision maker can develop additional alternatives to further investigate the base solution. The analysis shows that additional investments are essential since given the current infrastructure (i.e., desalination units, plants, distribution networks) only around 50% of the municipal water demand can be met. The analysis also provides the decision makers in Saudi Arabia with optimal operational values which maximize the flow while minimizing the total costs.
Conference Paper
The steady increase in oil prices and awareness regarding environmental risks due to carbon dioxide emissions are promoting the current interest in electric vehicles. However, the current relatively low driving range (autonomy) of these vehicles, especially compared with the autonomy of existing internal combustion vehicles, remains an obstacle to their development. In order to reassure a driver of an electric vehicle and allow him to reach his destinations beyond the battery capacity, we describe a system which generates an energy plan for the driver. We present in this paper the electric vehicle ecosystem and we focus on the contribution of using the generalized multi-commodity network flow (GMCNF) model as a vehicle routing model that considers energy consumption and charging time in order to ensure the usage of an electric vehicle beyond its embedded autonomy by selecting the best routes to reach the destination with minimal time and/or cost. We also present some perspectives related to the utilization of autonomous electric vehicles and wireless charging systems. We conclude with some open research questions.
In transition to a new era of human space exploration, the question is what the next-generation space logistics paradigm should be. The past studies on space logistics have been mainly focused on a "vehicle" perspective such as propulsive feasibility, cargo capacity constraints, and manifesting strategies, with the arbitrarily predetermined logistics network. But how do we select an optimal logistics network? Especially if we can utilize in-situ resources on the Moon and Mars, it will add complexity to network selection problem. The objective of this thesis is to develop a comprehensive graph-theoretic modeling framework to quantitatively evaluate and optimize space exploration logistics from a "network" perspective. In an attempt to create such a modeling framework, we develop a novel network flow model referred to as the generalized multi-commodity network flow (GMCNF) model. On top of the classical network flow problems, the GMCNF model proposed in this thesis introduces three types of matrix multiplications (requirement, transformation, and concurrency), and also allows loop edges associated with nodes (graph loops) and multiple edges between the same end nodes (multigraph). With this modification, the model can handle multiple commodities that interact with each other in the form of requirement at nodes, transformation on edges, and concurrency within edges. A linear programming (LP) formulation and a mixed integer linear programming (MILP) formulation of the GMCNF model are described in preparation for the two case studies. For the MILP formulation, in addition to the flow, we introduce two more variables, capacity expansion and decision binary, and additional constraints including the big-M method. The first case study applies the GMCNF LP model to human exploration of Mars. First we solve the baseline problem with a demand that is equivalent to that of the NASA's Mars Design Reference Architecture (DRA) 5.0 scenario. It is found that the solution saves 67.5% from the Mars DRA 5.0 reference scenario in terms of the initial mass in low-Earth orbit (IMLEO) primarily because chemical (LOX/LH2) propulsion is used along with oxygen-rich ISRU. We also present one possible scenario with two "gateway" resource depots at GTO and DTO with orbital transfer vehicles (OTVs) running in the cislunar and Martian systems. Then we solve variant problems that have different settings to see the effect of each factor. Findings include: taking advantage of oxygen-rich ISRU, LOX/LH2 is preferred to nuclear thermal rocket (NTR), the aerobraking option as well as ISRU availability on the Moon make great contributions in reducing the total mass to be launched from Earth, and as the ISRU production rate decreases, ISRU in each location becomes worthless at a certain threshold and the network topology changes toward direct paths using NTR. The other case study applies the GMCNF MILP model to the complex infrastructure systems in Saudi Arabia, focusing on the couplings between water and energy. Considering the capacity of the online infrastructures as of 2010 as a basis, we solve the problems with the 2030 demand and the 2050 demand. The objective function is a weighted sum of the total cost and the total CO2 emission. The key findings include: the network tends to be less connected, more isolated when putting more emphasis on minimizing the CO2 emissions, and some of the resulting networks suggest the possibility of the long-distance pipeline network connecting the west coast and the east coast via the central region (trans-peninsula pipeline).
Water is a constrained natural resource and in many areas of the planet water shortage is considered to be one of the most important issues to be resolved. This is certainly true for many Greek islands, where there is serious water shortage especially during the summer, thus hindering the development of the islands. The aim of the present work is to propose a method for the optimisation of water systems, i.e. the optimal water supply and distribution under conditions of water shortage, as they appear in the Aegean islands. In the water systems under consideration, the total demand as expressed by the users may exceed water availability. In this case, priorities between conflicting demands need to be taken into account. More specifically, the work describes the mathematical model that has been developed for the optimal allocation of water to various users from different sources with varying supply costs and water use values. Technical and environmental parameters are taken into account in the optimisation problem. Special emphasis is given to the implementation of the method in specific Aegean islands with water shortage. The novel feature of the work lies in the fact that it proposes an integrated framework for the solution of water resources optimisation, taking into account various problem parameters and thus resulting in important conclusions concerning supply sources, required infrastructure projects, water cost and value creating from the exploitation of water resources.
Appendix 8: Country case study – water policy reform in Saudi Arabia
  • H M H Sheikh
H.M.H. Al-Sheikh, Appendix 8: Country case study – water policy reform in Saudi Arabia, Proceedings of the second expert consultation on national water policy reform in the Near East (1997).
The Saudi Arabian Ministry of Water and Electricity annual technical report
  • Mowe
MoWE, The Saudi Arabian Ministry of Water and Electricity annual technical report (2013).
Interdependent Multicommodity Network Flow Modeling Framework
  • T Ishimatsu
  • O L De Weck
T. Ishimatsu, O.L. de Weck, Interdependent Multicommodity Network Flow Modeling Framework, Operations Research (2017). (Under review)
Country case study -water policy reform in Saudi Arabia
  • H M H Al-Sheikh
H.M.H. Al-Sheikh, Appendix 8: Country case study -water policy reform in Saudi Arabia, Proceedings of the second expert consultation on national water policy reform in the Near East (1997).