Page 5 - Reverse_osmosis2016

S. Casimiro et al. / Desalination and Water Treatment 61 (2017) 183–195 187

perspective. A conservative prediction might increase the regarding the reference value of 100% for the nominal heat
reliability of water production, but at the expense of less load output from the CSP plant (this is a user defined input),
efficient operation. Overall, the authors conclude that the with the aim of improving the MED plant performance
ROSA model provides an approximate estimate of sys- during part load operation of the CSP plant. This plant
tem performance that can be used in early stages of RO configuration favored the number of hours that the MED
system design. would operate at nominal capacity instead of optimizing
the average cutback that the MED plant would impose on
the electric production of the CSP plant. With that design,
4. Case study when the CSP exhaust steam heat load goes above the max-
imum heat absorption capacity of the MED plant (which in
4.1. System description the simulation from [2] was set to 40% of the nominal heat
The main goal of this study is to simulate the perfor- load output from the CSP), the remaining exhaust steam
mance of a parabolic trough plant coupled with a seawater is routed into the Sea Water Cooling Circuit (SWCC). The
desalination RO unit and compare it with an existing large- SWCC is set to operate at the same vapor pressure than the
scale (thermal vapor compression) TVC-MED parallel feed steam entering the MED plant. In these conditions, only
desalination plant, capable of producing 36 000 m d [8]. part of the exhaust steam is being used to power the MED
3
–1
This MED plant was chosen as reference for this case study, and produce fresh water. This produces an excessive cut-
firstly, because it is one of the few plants with detailed back on electric production due to the forced condensation
design information available in the literature. Secondly, at high pressure of the entire mass steam flow, and not only
because it was possible to use data from a previous study of the steam flowing into the MED. On the other hand this
regarding the operation of this MED plant using natural strategy ensures that the MED plant will operate more times
gas versus the option of using a CSP plant as power source during the year at nominal capacity. Because of these it was
[2]. Thirdly, no relevant detailed comparison is available in decided to change the CSP+MED system configuration and
the literature regarding the operation of CSP-RO, versus assume that the MED plant would be sized for this study
CSP-MED, versus the operation of an existing plant for the according to 100% of the heat load rejected by the CSP plant
same site. (1:1 ratio between CSP and MED). With this configuration
Such a coupling (CSP-RO) will initially assume that all a down condenser is considered only for the periods when
the net electrical power output from the power block will the CSP plant operates below the minimum load required
drive the RO unit’s high-pressure pump, pre and post-treat- by the MED, and for shutdown and startup procedures.
ment, intake and outfall systems. The unit’s main operat- To accommodate these changes to the CSP+MED con-
ing parameters, that is, the recovery and feed pressure, are figuration, the size of the CSP plant was reset to 49.4 MWe
established by considering membrane control and operation gross production and 44 MWe net, in the simulations shown
limits. The CSP-RO system modelled consists of a 49.4 MW in this paper—the minimum required to allow the opera-
e
–1
3
(gross production) parabolic trough CSP plant with a con- tion of a 36 000 m d MED plant in co-generation with a 1:1
ventional steam Rankine cycle coupled with a large-scale ratio. At design the CSP rejected heat load equals the MED
two-stage RO plant (first stage assumed to have 49 pressure required heat load (if the CSP installed capacity would not
vessels, and second stage 36 pressure vessels, each pressure have been reduced, the MED plant would need to be con-
–1
3
vessel with 6 elements). The RO system is divided into six sider larger than 36,000 m d , as it acts as the sole condens-
parallel connected trains, to enable flexible partial opera- er of the power cycle with the new system configuration
tion (each train with 2 stages). The RO system has a total using the 1:1 ratio). The size selected for the CSP plant is
recovery of 45%, and energy is recovered using a high effi- much larger than necessary for the RO system (~49.4 MWe
ciency pressure exchanger. The first stage recovers 37.6%, instead of ~6.7 MWe gross).
and the second stage 11.8% (the second stage receives as This study focuses especially on the water production
feed the brine produced on the first stage). Each simulated of the CSP-RO system using four different cooling systems
3
RO train produces 6 000 m d of fresh water, with a total with the CSP plant: Wet cooling (using fresh water), wet
–1
of 36,000 m d at nominal capacity (matching the output of cooling (using seawater), dry cooling and once-through sea-
3
–1
the full-scale MED plant described in [2]). water cooling assuming no grid connection in all cases. The
The CSP+MED system shown here is adapted from the location chosen for the system is the city of Trapani, in the
work presented in [2]. Since the release of the study per- southern island of Sicily, in Italy. The simulation for the CSP
formed in [2], the MED add-on to SAM has suffered contin- plant was done with the System Advisor Model’s (SAM)
uous upgrades. The simulations for the CSP+MED system (version 2014.1.14) physical trough model [9], using the
shown in this paper use an upgraded version of this add-on integrated TRNSYS software program. SAM is a validated
to SAM, that now makes use of new performance curves simulation program that can simulate the performance of
to simulate the Rankine cycle. This new upgrade allows CSP systems among other renewable energy systems using
the description of its operation with dedicated intermedi- hourly resource data. The simulations for the CSP+MED,
ate steam extractions to power the MED system (entering the once through and the seawater wet cooling systems are
either a thermal vapor compressor, TVC, and/or the ejec- performed using the add-on recently developed for SAM
tion system of non-condensable gases), though it will not be [2]. Since the publication of the work described in [2] the
described in detail in this article, as it will be part of a future upgrade performed to the add-on to SAM also includes a
publication. revised version of the once-through system, where now
In the original study [2], the CSP+MED system was the user can set either that the SWCC during operation will
simulated assuming that the MED system is undersized maintain a stable vapor pressure in the condenser, or that it