A Portable Hand-Operated Reverse Osmosis Desalination Device with Integrated Hydraulic Brine Energy Recovery

Abstract

Securing a freshwater supply for personnel engaged in remote maritime operations remains a persistent logistical challenge. This paper presents the design, hydraulic simulation, and experimental validation of a compact, manually operated seawater desalination device based on reverse osmosis (RO) coupled with an integrated hydraulic energy recovery system. The device employs a valve-commutated piston pump (cylinder bore 6 mm, rod diameter 3.7 mm, stroke 70 mm) driven by a lever-type handle mechanism. High-pressure brine rejected by the RO membrane is redirected via a two-position, three-way directional valve into the rod-end cavity of the pump cylinder, partially offsetting the filtration resistance and achieving an energy recovery ratio of 49.5%. Hydraulic circuit dynamics were analyzed using AMESim software, yielding a simulated freshwater output of approximately 0.02 L/min (1.2 L/h). A functional prototype with overall dimensions of 200 × 128 × 63 mm was fabricated and tested under 3.57% salinity conditions. Five consecutive trials produced a mean freshwater flow rate of approximately 1.15 L/h (desalination rate exceeding 95%), confirming consistency with the simulation predictions and satisfying the design requirements for individual field use in remote maritime settings.

1. Introduction

Access to potable water in remote maritime environments is a critical and recurring challenge for offshore workers, fishing crews, maritime search-and-rescue teams, and isolated island communities during disaster-induced water shortages [1,2]. In such settings, conventional freshwater resupply is logistically expensive, slow, or entirely impractical, and the lack of a reliable electrical power supply further constrains the use of conventional purification equipment. On-site seawater desalination therefore offers an attractive self-sufficient solution, yet existing desalination technologies vary enormously in scale, cost, and energy demand, and only a narrow subset is suited to truly portable, off-grid use [3]. Within the broader context of sustainable water–energy systems and the transition toward carbon-neutral technologies, off-grid desalination approaches that minimise external energy input are attracting increasing attention as environmentally responsible solutions for decentralised water supply [4].

Among desalination methods, reverse osmosis (RO) has become the dominant technology worldwide because of its comparatively low energy consumption relative to thermal distillation, its inherent modularity, and the steady improvement in membrane permeability and salt rejection [5,6]. The first asymmetric cellulose-acetate RO membranes were developed by Loeb and Sourirajan, but their water permeability was only about 0.2 L·m⁻²·h⁻¹·bar⁻¹ at 55 bar [7]. The introduction of thin-film composite (TFC) polyamide membranes—formed by interfacial polymerization of m-phenylenediamine and trimesoyl chloride on a polysulfone support—raised permeability to roughly 1.2 L·m⁻²·h⁻¹·bar⁻¹ with NaCl rejection exceeding 99.4% [8]. Subsequent advances in substrate nanostructure, monomer chemistry, and surface functionalization have pushed commercial seawater RO (SWRO) membranes beyond 2.0 L·m⁻²·h⁻¹·bar⁻¹ at 99.7% rejection [9], while emerging nanocomposite membranes incorporating graphene oxide, zeolites, or carbon nanotubes have demonstrated flux gains of up to 80% at comparable rejection, albeit with persisting manufacturing challenges [10]. These developments have made compact RO modules increasingly viable for small-scale applications.

A key determinant of RO energy efficiency is the recovery of hydraulic energy from the rejected brine. In a typical SWRO process, the high-pressure brine leaving the membrane retains 60–70% of the energy supplied to the feed pump; if discarded, this energy is entirely lost, inflating operating costs by 30–50% [11]. To address this, large-scale plants routinely employ energy recovery devices (ERDs) that capture the brine pressure and return it to the feed stream, enabling specific energy consumptions below 3.5 kWh·m⁻³ [12]. Representative ERD architectures include isobaric pressure exchangers, work exchangers such as the DWEER family, and turbine-based systems, with reported recovery efficiencies of 93–98% [13,14]. Numerous refinements have followed: piston- and rotary-type ERDs minimising seawater–brine mixing [15], reciprocating fluid-switching designs exceeding 96% efficiency [16], and fully rotary-valve devices reaching 98% with leakage scaling as the cube of the clearance gap [17]. However, all of these devices are conceived for fixed installations or medium-to-large throughputs, and their size, complexity, and reliance on high-precision rotating components make them fundamentally unsuitable for hand-held use.

For portable, individual-use applications, an entirely different design philosophy is required: the unit must be lightweight and ideally hand-held, mechanically simple, operable without any electrical power, and capable of producing sufficient freshwater to sustain a person. Existing portable systems illustrate the trade-offs involved. The Swiss Katadyn Survivor series represents the current commercial benchmark; the Survivor 06 achieves a salt rejection of 98.4% but delivers only about 0.5 L·h⁻¹ under manual operation and incorporates no active energy recovery [18]. At the opposite extreme, the U.S. ROWPU-class systems offer high throughput but require tens of kilowatts of generated power and truck-scale logistics, disqualifying them from individual carriage [19]. Research-stage efforts have explored solar-powered portable units, whose output is constrained by weather and whose electrical components are vulnerable to seawater corrosion [20], and compact powered devices that integrate a pump and electronic control within a portable housing but rely on an onboard power source and incorporate no brine energy-recovery mechanism [21].

This brief survey reveals a clear gap. Although both RO membrane technology and brine energy recovery are individually mature, no published work simultaneously delivers (i) a hand-held, purely manual form factor, (ii) an integrated brine energy-recovery function, and (iii) a closed-loop study spanning hydraulic design, dynamic simulation, and experimental prototype validation. Existing portable devices either omit energy recovery entirely or rely on electrical actuation that is impractical and corrosion-prone in marine environments.

The present work addresses this gap with a fully manual, hand-held RO desalination device that integrates a passive, mechanically synchronized brine energy-recovery circuit. The central innovation lies in repurposing a two-position, three-way (2/3) directional valve—mechanically coupled to the pump piston through a lever handle—to redirect high-pressure reject brine into the rod-end cavity of the pump cylinder, thereby reusing the brine energy to assist each delivery stroke without any sensors, electronics, or auxiliary machinery. The specific contributions of this study are: (i) a compact hydraulic-circuit design coupling a valve-commutated piston pump with a 2/3 directional valve for passive brine energy recovery; (ii) a parametric kinematic model enabling AMESim simulation of the manual pumping dynamics; (iii) quantitative simulation of flow rates, piston forces, and the resulting energy-saving ratio; and (iv) fabrication and experimental validation of a functional prototype under controlled ocean-salinity conditions. The remainder of this paper is organized as follows: Section 2 details the device design and structural modelling; Section 3 presents the hydraulic-circuit simulation, prototype fabrication, and performance testing; and Section 4 concludes the study.

2. Materials and Methods

2.1. Operating Principle and Passive Brine Energy-Recovery Concept

The proposed device integrates three functional subsystems within a single hand-held housing: a valve-commutated piston pump that draws raw seawater and delivers it at the required pressure to the membrane module, a reverse-osmosis (RO) membrane module that separates the pressurized feed into low-pressure permeate and high-pressure reject brine, and a passive energy-recovery circuit built around a two-position, three-way (2/3) directional valve that is mechanically coupled to the pump piston. The defining feature of the device is its passive, mechanically synchronized brine energy-recovery mechanism, which distinguishes it from both the electrically actuated energy recovery devices used in fixed installations and the energy-recovery-free portable units reviewed in Section 1. In contrast to rotary or pressure-exchanger devices that rely on precision rotating assemblies or sensor-driven actuation, the present mechanism reuses the brine pressure entirely through the kinematic coupling between the piston rod and the directional-valve spool and therefore requires no electrical power, no sensors, and no auxiliary machinery.

The operating cycle consists of an intake stroke and a delivery stroke. During the intake stroke, the operator pulls the lever, the piston retracts, and seawater is drawn into the blind-end (rod-free) cavity through the inlet check valve; at the same time, the lever displaces the 2/3 valve spool to its left position so that the rod-end cavity is connected to the brine drain and any residual brine is vented. During the delivery stroke, the operator pushes the lever, the piston advances, and the seawater contained in the blind-end cavity is forced through the outlet check valve into the RO membrane at the operating pressure; concurrently, the lever switches the valve spool to its right position, routing the high-pressure reject brine from the membrane into the rod-end cavity, where it acts on the annular face of the piston and produces an assistive force that partially offsets the manual effort required to maintain the membrane operating pressure. Because the valve switching is rigidly tied to the stroke direction, energy recovery occurs automatically within every delivery stroke without any external control. The quantitative basis of this mechanism is formalized in the following subsections.

2.2. Force Balance and Energy-Recovery Model

The energy-recovery behaviour of the device can be described by a force balance on the piston during the delivery stroke. Let $p_m$ denote the RO membrane operating pressure acting on the blind-end face of area $A_1$, and $p_b$ the recovered brine pressure acting on the annular rod-end face of area $A_2$. The two effective areas are determined by the cylinder bore $D$and the rod diameter $d$as

$$A_1={\displaystyle \frac{\pi D^2}{4}},\hspace{1em}\hspace{1em}{A}_2={\displaystyle \frac{\pi (D^2-d^2)}{4}}$$

(1)

Following standard hydraulic-cylinder force analysis [22], in the absence of energy recovery, the axial force that the operator must supply to drive the delivery stroke is governed solely by the membrane back-pressure together with a lumped term $F_f$ representing friction and check-valve cracking losses,

$$F_{\mathrm{conv}}=p_m{A}_1+F_f$$

(2)

When the passive recovery circuit is engaged, the recovered brine pressure contributes an assistive force on the annular face, so that the net operator force is reduced to

$$F_{\mathrm{push}}=p_m{A}_1-p_b{A}_2+F_f$$

(3)

The brine energy reclaimed and reused per delivery stroke of length Lis $E_{\mathrm{rec}} =p_b$ $A_{2}L$. Defining the brine energy-recovery ratio ${\eta }_{\mathrm{rec}}$ as the fraction of operator effort offset by the recovered brine, and noting that throughout a steady RO process the reject brine retains nearly the full feed pressure so that $p_b\approx p_m$, the recovery ratio can be expressed as

$${\eta }_{\mathrm{rec}}={\displaystyle \frac{F_{\mathrm{conv}}-F_{\mathrm{push}}}{F_{\mathrm{conv}}}}\approx {\displaystyle \frac{p_b{A}_2}{p_m{A}_1}}=\left({\displaystyle \frac{p_b}{p_m}}\right)\left(1-{\displaystyle \frac{d^2}{D^2}}\right)$$

(4)

Equation (4) constitutes the central design relation of the device, since it shows that the achievable energy-recovery ratio is governed almost entirely by the rod-to-bore diameter ratio $d/D$. A larger rod recovers a greater fraction of the brine energy and thus yields a higher ${\eta }_{\mathrm{rec}}$, but it simultaneously reduces the net permeate volume delivered per stroke; conversely, a smaller rod maximises permeate output at the cost of reduced energy recovery. The selection of cylinder geometry must therefore reconcile these two competing objectives, a trade-off that is made explicit in the sizing procedure of the following subsection.

2.3. Pump Cylinder Sizing

The net freshwater delivered per stroke equals the difference between the volume displaced by the blind-end face and the volume occupied by the rod within the rod-end cavity. Assuming equal intake and delivery velocities ($v_1=v_2=v$) and a volumetric efficiency ${\eta }_v$, the blind-end and rod-end flow rates are

$$q_1={\displaystyle \frac{v\pi D^2}{4{\eta }_v}},\hspace{1em}\hspace{1em}{q}_2={\displaystyle \frac{v\pi (D^2-d^2)}{4{\eta }_v}}$$

(5)

so that the net permeate flow rate depends only on the rod diameter,

$$q=q_1-q_2={\displaystyle \frac{v\pi d^2}{4{\eta }_v}}$$

(6)

Imposing the design target $q=3.33\times {10}^{-7} {\mathrm{m}}^3\hspace{0.17em}{\mathrm{s}}^{-1}$ (equivalent to 1.2 L·h⁻¹) together with $v=0.03 {\mathrm{m}·\mathrm{s}}^{-1}$ and ${\eta }_v=0.96$ yields a required rod diameter of $d=3.7 \mathrm{mm}$. A bore-to-rod ratio of approximately 1.6, corresponding to a bore of $D=6 \mathrm{mm}$, was then adopted to satisfy two simultaneous requirements. First, substituting this ratio into Equation (4) gives a theoretical energy-recovery ratio of ${\eta }_{\mathrm{rec}}\approx 1-(3.7/6{)}^2\approx 0.62$, confirming that the chosen geometry provides a substantial brine-assist contribution. Second, the resulting swept volume remains sufficiently small to preserve the compact, hand-held form factor of the device. The stroke length was set to $L=70 \mathrm{mm}$, giving a full-cycle piston travel of 140 mm. These parameters were carried forward without modification into the AMESim model and into the prototype reported in Section 3.

2.4. Two-Position, Three-Way Directional Valve

The 2/3 directional valve is the keystone of the energy-recovery circuit. Its spool features two land diameters slightly larger than the corresponding port diameters, with a recommended diametric clearance of 0.005–0.04 mm in order to minimize internal leakage [22]. Port A is connected to the high-pressure side of the circuit, port P to the rod-end cavity, and port T to the brine drain. In the left-position state, which corresponds to the intake stroke, port A communicates with port T, venting any residual brine from the rod-end cavity; in the right-position state, which corresponds to the delivery stroke, port P communicates with port A, introducing high-pressure brine into the rod-end cavity so that it assists the advancing piston. The spool is mechanically linked to the piston rod through the lever handle, so that valve commutation is intrinsically synchronized with the stroke direction. This kinematic coupling is the structural embodiment of the passive recovery concept defined in Section 2.2, as it removes the need for the timing sensors, solenoids, and control electronics that conventional switching energy recovery devices require, and it thereby ensures that the recovery function remains entirely self-actuated and robust in the seawater environment.

2.5. Lever Handle Mechanism

A lever-type handle is employed both to reduce the operator effort and to provide the single mechanical input that synchronously actuates the piston and the valve spool. The mechanism comprises four kinematic links, namely the handle, a slider coupled to the piston rod, the piston rod itself, and the fixed pivot, which together form two sliding pairs and two revolute pairs. The mobility of the mechanism is evaluated using the Grübler–Kutzbach criterion for planar linkages [23],

$$F=3n-2P_L-P_H=3\left(3\right)-2\left(4\right)-0=1$$

(7)

The result confirms that the mechanism possesses a single degree of freedom and therefore executes a unique, well-defined motion under a single input. With a handle arm of 200 mm measured from the pivot to the grip and a piston-connection point located 35 mm from the pivot (Figure 1), the lever provides a mechanical advantage of $i=200/35\approx 5.7$. The device consequently achieves a two-stage reduction in operator effort: the passive brine recovery first lowers the axial piston force from $F_{\mathrm{conv}}$ to $F_{\mathrm{push}}$ in accordance with Equation (3), and the lever subsequently divides this force by the mechanical advantage $i$, so that the effective grip force becomes $F_{\mathrm{grip}}=F_{\mathrm{push}}/i$. This compounded reduction, in which energy recovery acts in series with mechanical amplification, is what renders sustained manual operation feasible against a membrane back-pressure of approximately 50 bar.

2.6. RO Membrane Module and Overall Configuration

A single-stage (single-pass) spiral-wound thin-film composite (TFC) polyamide RO membrane element, of the seawater-desalination grade, was selected for the device and housed within a cylindrical canister fitted with a threaded end-cap to permit field replacement. A standard seawater RO operating pressure of approximately 50 bar was adopted as the hydraulic design pressure of the circuit. The membrane assembly is positioned beneath the pump and valve unit, with separate outlet lines for permeate and brine concentrate, while a single inlet line feeds raw seawater to the pump. The complete assembly, comprising the valve-commutated pump, the 2/3 directional valve, the lever handle, and the membrane canister, occupies overall package dimensions of 200 × 128 × 63 mm and constitutes a self-contained, hand-held unit suitable for off-grid operation.

2.7. Research Methodology

The work proceeds through four sequential stages. In the design stage, the functional requirements of portability, manual off-grid operation, and brine energy recovery are translated into the component-level design of the valve-commutated piston pump, the 2/3 directional valve, the lever handle, and the membrane module. In the modelling stage, the theoretical force-balance and sizing relations of Equations (1)–(7) are established and complemented by two- and three-dimensional geometric models of the mechanism. In the simulation stage, a complete hydraulic-circuit model is constructed in AMESim and exercised to predict the flow rates, piston forces, and energy-saving behaviour of the device. Finally, in the validation stage, a functional prototype is fabricated and tested under controlled ocean-salinity conditions to measure the freshwater flow rate and desalination rate. The four stages are closed into a loop by the mutual cross-validation between the simulation predictions and the experimental measurements, which ensures the internal consistency of the design methodology.

3. Results and Discussion

3.1. Simulation Platform and Model Architecture

Siemens AMESim (Simcenter Amesim, version 2021; Siemens Digital Industries Software, Plano, TX, USA) was selected as the simulation platform on account of its mature hydraulic component libraries, robust numerical solvers, and capacity for mixed mechanical–hydraulic–signal modelling within a single environment. AMESim has been widely employed for the dynamic analysis of water-hydraulic systems and small-scale desalination circuits, and its validated check-valve, directional-valve, and piston-cylinder models are directly applicable to the present device.

The complete hydraulic circuit was assembled from four standard AMESim libraries: the Hydraulic library (fluid power components), the 1D Mechanical library (rigid-body kinematics), the Hydraulic Component Design (HCD) library (parametric piston elements), and the Signal library (control signals and data acquisition). The resulting model, shown in Figure 2, is organized into two functionally distinct sub-systems: (i) the valve-commutated plunger-pump assembly together with its swash-plate drive mechanism; and (ii) the hydraulic energy recovery circuit centred on the two-position, three-way (2/3) directional valve.

The overall circuit logic follows the operating sequence established by the hydraulic schematic: during the intake stroke the directional valve is in the left position, the inlet check valve opens and raw seawater fills the blind-end cavity while brine in the rod-end drains through the T port; during the delivery stroke the valve shifts to the right position, the outlet check valve opens, seawater is pressurized and delivered to the RO membrane, and high-pressure brine from the membrane is simultaneously admitted into the rod-end cavity through the P port to assist the piston.

3.2. Pump Sub-System Modelling

The physically manufactured pump is a valve-commutated piston pump. For the purpose of simulating steady-state freshwater production characteristics, the pump is modelled in AMESim as a valve-commutated plunger pump, which shares the same operating principle—periodic blind-end filling and expulsion controlled by check valves—while offering a numerically cleaner representation of the piston-cylinder interface. This modelling choice does not affect the flow-rate, pressure, or force results that are of primary interest in this study.

The pump cylinder is represented by two coupled HCD piston-cylinder elements: a blind-end (rod-free) element and a rod-end element, whose geometric parameters are configured to reflect the physical dimensions of the fabricated pump cylinder, as detailed in

Table 1. AMESim parameter settings for the pump piston-cylinder elements.

Parameter	Blind-End Cavity	Rod-End Cavity	Unit
Piston (bore) diameter, D	6	6	mm
Rod diameter, d	0 (no rod)	3.7	mm
Maximum stroke, L	70	70	mm
Dead-band (start-up offset)	1	1	mm
Hydraulic fluid index	Seawater	Seawater	—

below.

A variable-volume chamber element is connected in parallel with the blind-end port, with its minimum volume set to approximately 1% of the blind-end swept volume. This element serves as a numerical compliance buffer, preventing pressure singularities at stroke reversal when the instantaneous volumetric flow transitions from inflow to outflow.

In addition, the manually operated piston rod is modelled by an equivalent swash-plate mechanism—a standard representation in AMESim for axial-piston machines—in which a rotary motor drives a swash plate that converts rotational motion into the reciprocating axial displacement of the plunger. The equivalence is exact for sinusoidal piston kinematics. Figure 3 illustrates the geometric derivation of the plunger displacement law.

With reference to Figure 3, when the swash plate has rotated through angle θ, its radius R projects onto the radial perpendicular plane as:

$$l=R\cdot cos\theta$$

(8)

The axial component of this projection, defined as the plunger displacement measured from the top-dead-centre position, is:

$$l^{\prime }=-l\sin \alpha =-R\sin \alpha \cos \theta$$

(9)

where α is the fixed swash-plate inclination angle (α = 30° in the present model). From this equation, the maximum effective stroke is:

$$L=2R\sin \alpha$$

(10)

Setting L = 70 mm and α = 30° gives R = L/(2 sin 30°) = 70/(2 × 0.5) = 70 mm. This value of R = 70 mm (represented as 0.07 m in SI units) is entered as the gain factor in the AMESim formula block. The swash-plate inclination α = 30° is the constant input to the formula block. The angle signal θ (in degrees converted to radians within the block) is provided by an angle sensor connected to the rotary motor output. The formula entered into the AMESim expression editor is:

$$\mathrm{Output}=-\sin (x\cdot \pi /180)\cos (y\cdot \pi /180)$$

(11)

where x is the inclination angle input (30°) and y is the real-time swash-plate rotation angle θ from the sensor. The product of this formula and the gain R yields the time-varying plunger displacement l′(t). The rotary motor speed is set to 6 rev/min. Given the swash-plate kinematic relationship, this yields a peak plunger velocity of 0.044 m/s. Since the plunger follows a sinusoidal velocity profile, its RMS velocity is equal to the peak value divided by √2, giving an RMS velocity of approximately 0.031 m/s, which closely matches the manual operation target of 0.03 m/s. The RMS velocity, rather than the peak velocity, is used as the matching criterion because it represents the effective time-averaged velocity over a complete stroke cycle and therefore provides a more physically meaningful basis for comparing the simulated pumping effort with the sustained hand-operation condition. Figure 4 shows the resulting plunger displacement and velocity traces over 30 s of simulation (three complete cycles at 0.1 Hz). The velocity curve is the corresponding negative sine wave with peak magnitude 0.044 m/s.

3.3. Energy Recovery Sub-System Modelling

The energy recovery sub-system is composed of four AMESim elements: the 2/3 directional valve, a square-wave signal source, a hydraulic cylinder representing the brine concentrate accumulation vessel, and a relief valve representing the RO membrane module.

The 2/3 directional valve is modelled as a solenoid-actuated proportional valve with a rated current of 40 mA. Since the overall device is very compact and flow velocities through the valve are low, pressure-drop effects across the valve are negligible and are not modelled. The valve parameter settings are listed in

Table 2. AMESim parameter settings for the 2/3 directional valve.

Parameter	Value	Unit
Hydraulic fluid index	Seawater (index 0)	—
Valve rated current	40	mA
Working mode at 40 mA	Right position (P → A, delivery stroke)	—
Working mode at 0 mA	Left position (A → T, intake stroke)	—
Pressure-drop model	Neglected (compact device)	—

.

Valve switching is controlled by a square-wave signal source with the parameters listed in

Table 3. AMESim square-wave signal source parameters (valve control).

Parameter	Value	Unit/Type
Transition type	Discontinuous	—
Square-wave frequency	0.1	Hz
Pulse ratio (duty cycle)	50	% (null units)
Minimum level (low state)	0	mA (null units)
Maximum level (high state)	40	mA (null units)
Phase delay	0	s

. The signal transitions between 0 mA (left position, intake stroke) and 40 mA (right position, delivery stroke) at a frequency of 0.1 Hz, precisely synchronized with the plunger displacement curve: when the plunger displacement l′ is positive (intake half-cycle), the signal output is 0; when l′ is negative (delivery half-cycle), the output is 40. This synchronization is achieved by matching the square-wave frequency and phase to the rotary motor speed at a frequency of 0.1 Hz, synchronized with the plunger displacement (0 mA during intake, 40 mA during delivery).

The RO membrane module is represented in AMESim by a pilot-operated relief valve (safety valve) with a cracking pressure of 50 bar, consistent with the design working pressure of the device. The relief valve serves a dual purpose: it imposes the required back-pressure on the high-pressure seawater delivered by the pump, thereby simulating the osmotic and hydraulic resistance of the membrane; and it provides overpressure protection by diverting excess flow back to the brine storage cylinder if the system pressure exceeds 50 bar during transient events. The relief valve parameters used in the simulation are summarized in

Table 4. AMESim relief valve (RO membrane surrogate) parameter settings.

Parameter	Value	Unit
Hydraulic fluid index	Seawater	—
Cracking pressure (membrane operating pressure)	50	bar
Flow rate–pressure gradient	500	L/min/bar
Valve hysteresis	0	bar

.

A separate hydraulic cylinder element is included in the brine circuit to model the physical volume of concentrate held in the brine discharge line. During the intake stroke, this cylinder accepts the brine volume expelled from the rod-end cavity through the T port; during the delivery stroke, it supplies brine to the rod-end cavity through the P port. The brine cylinder does not represent an additional energy storage device; it serves purely as a lumped-volume element to balance the brine mass in the circuit.

3.4. Simulation Results and Analysis

Because the device is intended for seawater desalination, the working fluid throughout the entire AMESim model—including both the raw seawater side and the brine side—is declared as seawater rather than the default mineral hydraulic oil. The fluid properties were assigned using AMESim’s fluid declaration element and are listed in

Table 5. Working fluid properties declared in AMESim (seawater at 23.2 °C).

Property	Value	Unit
Density, ρ	1020	kg/m3
Bulk modulus, β	23,500	bar
Absolute (dynamic) viscosity, μ	5.49 × 10−4	Pa·s

. The values correspond to standard ocean water at the experimental ambient temperature of 23.2 °C.

The declared bulk modulus of 23,500 bar is characteristic of seawater at moderate pressure and is considerably higher than that of mineral oil (~14,000 bar), which increases the hydraulic stiffness of the modelled system and reduces pressure transients at valve switching events. The viscosity of 5.49 × 10⁻⁴ Pa·s is substantially lower than that of typical mineral hydraulic oil, which lowers viscous friction losses in the valve spool and check-valve spring elements.

The simulation was run for a total duration of 30 s at an integration step of 1 ms, covering exactly three complete pumping cycles (one cycle = 10 s at 0.1 Hz). All results presented below are taken from steady-state operation; the first 0–2 s transient associated with hydraulic system initialization is excluded from quantitative analysis. Key simulation results and their comparison with design targets are summarized in

Table 6. Summary of key AMESim simulation results compared with design targets.

Performance Metric	Simulation Result	Design Target	Unit
Blind-end peak flow rate	±0.085	—	L/min
Rod-end peak flow rate	±0.047	—	L/min
Net freshwater peak flow rate	0.028	—	L/min
Net freshwater RMS flow rate	0.020	0.020	L/min
Freshwater hourly production	1.20	1.20	L/h
Peak piston force—intake (pull)	141	—	N
RMS piston force—intake (pull), F1	99	—	N
Peak piston force—delivery (push)	72	—	N
RMS piston force—delivery (push), F2	50	—	N
Energy saving ratio (brine assistance)	49.5%	>40%	—

.

The 49.5% energy-saving ratio reported above is derived from the AMESim simulation, in which the directional valve is treated as leak-free and the hydraulic connections as loss-free. In the physical device, several non-ideal effects would reduce the actually attainable recovery ratio: finite spool clearance in the 2/3 directional valve permits a small cross-port leakage of high-pressure brine; hydraulic losses arise in the connecting passages and at the valve ports; and mechanical friction occurs at the piston seals, the lever pivot, and the sliding pairs. Each of these dissipates a fraction of the recoverable brine energy, so that the effective in-service energy-saving ratio is expected to be somewhat lower than the simulated value. A direct experimental determination of the recovered force—by instrumenting the prototype with an in-line force sensor on the piston rod—is therefore identified as a priority for future validation of the energy-recovery performance.

Figure 5 shows the flow rate at the blind-end cavity port as a function of time. The waveform is quasi-cosinusoidal with peak magnitude approximately ±0.085 L/min and period 10 s, consistent with the swash-plate kinematic model. Positive flow corresponds to the intake stroke (seawater inflow into the blind end); negative flow corresponds to the delivery stroke (seawater outflow). The waveform is symmetric about zero, confirming that the intake and delivery stroke volumes are equal, as required by the constant-velocity pumping assumption.

The two check valves correctly separate the intake and delivery flows, and the mass balance across the pump is preserved throughout the cycle.

Figure 6 shows the rod-end cavity flow rate. During the delivery stroke, high-pressure brine is admitted into the rod-end through the P port of the 2/3 directional valve, producing a positive flow peak of approximately 0.047 L/min. During the intake stroke, the residual brine in the rod-end is expelled through the T port (brine drain line), producing a negative flow trough of the same magnitude. The waveform is again quasi-sinusoidal.

Figure 7 presents the overlay of the blind-end (seawater) and rod-end (brine) flow-rate curves. During the delivery stroke, both the outward seawater flow at the blind end and the inward brine flow at the rod end occur simultaneously, consistent with the device operating principle: the pump pressurizes seawater toward the membrane while brine pressure assists the piston from the opposite face. The difference between their peak magnitudes—approximately 0.038 L/min—represents the net volume displaced by the piston rod per unit time, which corresponds to the theoretical maximum freshwater production rate per cycle. Taking the RMS value of this difference over a complete stroke cycle yields a net freshwater flow rate of approximately 0.020 L/min, equivalent to 1.20 L/h. This result agrees exactly with the analytical design prediction, confirming internal consistency between the hydraulic sizing model and the dynamic simulation.

The 2/3 directional valve correctly routes the brine to the rod-end cavity during delivery and to the drain during intake, with no cross-port leakage in the simulation.

During the intake stroke (signal = 0, left position), port P is blocked and port A communicates with port T: brine from the rod-end cavity flows outward through A and drains via T. During the delivery stroke (signal = 40, right position), port T is blocked and port A communicates with port P: high-pressure brine from the membrane brine outlet flows inward through P and into the rod-end cavity via A. This behaviour is confirmed by the simulation: port P exhibits zero flow during the intake half-cycle and non-zero flow during the delivery half-cycle, while port T shows the opposite pattern. Port A alternates accordingly.

The peak flow rate at each of the three ports is approximately 0.046–0.047 L/min, which is equal to the rod-end cavity peak flow. This equality is expected because the directional valve is the sole pathway for brine to enter and exit the rod-end, and any pressure-drop effect is negligible at this scale. The equality of peak flow rates across all three ports additionally confirms that the valve introduces no internal leakage or bypass in the simulation, consistent with the zero-clearance spool model used.

The brief flow impulses visible at the switching instants (t = 5, 10, 15, 20, 25 s) are a numerical artefact caused by the discontinuous step transition of the square-wave signal. In the instantaneous moment of valve switching, the rod-end cavity volume is transiently disconnected from both the P and T supply lines, causing a pressure equalization pulse. The duration of these pulses is less than 0.05 s (five integration steps), and their integrated volume is below 0.001 mL per event—negligible relative to the 4.7 mL stroke volume. They have no physical counterpart in the actual manually switched device, where valve switching is gradual and mechanically constrained.

The algebraic relationship that governs the mass balance at the membrane node is:

$$q_{blind,out}=q_{membrane}+q_{rod,in}$$

(12)

where q_blind,out is the seawater volumetric outflow from the blind end (delivery stroke), q_membrane is the flow through the relief valve (membrane permeate + concentrate, treated here as the concentrate bypass), and q_rod,in is the brine inflow to the rod-end. From the simulation: the sum of the relief valve peak (0.072 L/min) and the rod-end peak (0.047 L/min) is 0.119 L/min at t = 7.5 s, the apparent difference from the blind-end outflow peak of 0.085 L/min arises from the phase offset between the individual peaks (which do not occur at the same instant) rather than from any violation of mass conservation, as confirmed by the instantaneous balance holding at each time step.

Figure 8 shows the net axial force acting on the piston rod as a function of time, which is equivalent to the manual push/pull force required of the operator. The simulation yields a peak intake pull force of 141 N and an RMS value of approximately 99 N, while the delivery stroke requires a peak push force of only 72 N with an RMS value of approximately 50 N.

The energy consumed per stroke is proportional to the product of the force and the stroke length $l’$. The energy required for the intake stroke is $W_1= F_1 \times l’$, and for the delivery stroke $W_2 = F_2 \times l’$. The fractional energy saving due to the brine-assist mechanism is therefore:

$${\eta }_{save}=\left(\begin{array}{c}{W}_1-W_2\end{array}\right)/W_1=\left(\begin{array}{c}{F}_1-F_2\end{array}\right)/F_1=\left(\begin{array}{c}99-50\end{array}\right)/99\approx 49.5\%$$

(13)

The physical origin of this force reduction is the high-pressure brine acting on the annular rod-end face of the piston during the delivery stroke. At the membrane operating pressure of 50 bar, the brine exerts a passive assistive force on the rod-end annular face that partially counteracts the resistance encountered during seawater pressurization, thereby reducing the net push force required of the operator. The analytical estimate of this assistive force is consistent with the simulated peak force reduction of approximately 69 N (from a peak pull force of 141 N during intake to a peak push force of 72 N during delivery); the recovered force does not reach the full theoretical value because the brine pressure in the rod-end cavity varies dynamically throughout the delivery stroke and the full membrane operating pressure of 50 bar is not sustained over the entire half-cycle.

Considering the combined effect of brine-assisted energy recovery and the lever mechanical advantage, the operator experience is considerably more manageable than the raw piston forces suggest. Without the lever, the intake stroke would require a peak pull force of 141 N, which is at the upper limit of comfortable sustained one-handed operation. With the 5.7:1 lever mechanical advantage, the equivalent grip force at the handle tip is reduced to approximately 25 N during the intake stroke and approximately 13 N during the delivery stroke—both well within the range of comfortable, prolonged manual operation—confirming that the device can be operated continuously without undue fatigue.

3.5. Prototype Fabrication and Performance Testing

A fully functional prototype was fabricated to the dimensions (200 × 128 × 63 mm). The prototype is shown in Figure 9. The assembly is organized vertically into three structural sections: the upper lever handle provides the actuating moment; the middle trapezoidal body houses the piston pump above and the directional valve below; and the lower cylindrical canister contains the RO membrane module.

Table 7. Key fabrication parameters of the prototype.

Component/Parameter	Specification	Unit
Overall dimensions (L × H × W)	200 × 128 × 63	mm
Total assembled mass (without water)	~450	g
Pump body and valve housing material	Aluminium alloy 6061-T6	—
Lever frame material	Aluminium alloy 6061-T6	—
Membrane canister material	Acrylic (PMMA) tube, stainless-steel end fittings	—
RO membrane type	Single-stage spiral-wound TFC polyamide, seawater grade	—
RO membrane design operating pressure	50	bar
Piston bore diameter, D	6	mm
Piston rod diameter, d	3.7	mm
Piston stroke, L	70	mm
Lever arm (pivot to grip)	200	mm
Lever arm (pivot to piston connection)	35	mm
Lever mechanical advantage	5.7:1	—

lists the key geometric and material parameters of the fabricated prototype.

Because the experiments were conducted under controlled laboratory conditions, authentic ocean water was not available; artificial seawater was prepared by dissolving instant marine sea-salt crystals in deionized water. The target salinity was 3.5% (w/w), representing typical open-ocean salinity (33–35 g/L NaCl equivalent).

Preparation procedure: a known mass of sea-salt crystals was added to deionized water in a 2 L graduated beaker and stirred until fully dissolved. Salinity was measured using a handheld ion (salinity) meter, which reads salinity and temperature simultaneously. The salt-to-water ratio was iteratively adjusted until the measured salinity stabilized at 3.57 ± 0.05% at the experimental ambient temperature of 23.2 °C (296.4 K). This value was recorded as the feed salinity, C_feed = 3.57%, for all five experimental trials.

The choice of 3.57% salinity—slightly above the average ocean salinity of 3.5%—is deliberate: it represents a moderately challenging feed condition that provides a conservative (slightly pessimistic) benchmark for the desalination rate, since a higher salt concentration increases the osmotic pressure and therefore the energy required for separation. The experimental setup is shown in Figure 10, comprising the prototype device and three vessels: a feed bucket (right) containing the artificial seawater feed at 3.57% salinity, into which the prototype inlet tube is fully submerged; a brine discharge bucket (left) collecting the concentrated brine expelled from the device without recirculation; and a 200 mL graduated glass beaker on the bench collecting the product water from the freshwater outlet tube.

Before each trial, both collection vessels were emptied and dried, the feed salinity was confirmed within 3.57 ± 0.05%, and the product water beaker was tared to zero. Each trial consisted of 120 continuous manual pump strokes at a steady cadence of 1 Hz over 2 min, maintaining a consistent grip force and stroke amplitude throughout. At the end of each trial, the accumulated product volume was read from the graduated beaker to the nearest 2 mL graduation, and the product salinity was measured with the handheld salinity meter, which was rinsed with deionized water between measurements. The beaker was then emptied and dried, and the feed salinity was re-confirmed before proceeding to the next trial. Five consecutive trials were conducted (Trials 1–5) using the same feed solution. The freshwater production rate (flow rate) was calculated from the accumulated volume and the trial duration:

$$q_{fw}=V_{product}/t\times 60$$

(14)

where V_product is the measured product volume and t = 2 min is the trial duration. The desalination rate (salt rejection) was calculated as:

$$DR=\left(\begin{array}{c}{C}_{feed}-C_{product}\end{array}\right)/C_{feed}\times 100\%$$

(15)

where C_feed = 3.57% is the measured feed salinity and C_product is the measured product salinity for that trial. The measured results for all five trials are presented in

Table 8. Experimental performance results for all five trials.

Trial	Product Volume, V (mL)	Flow Rate, q_fw (mL/h)	Product Salinity, C_product (%)	Desalination Rate, DR (%)
1	38	1140	0.352	90.14
2	39	1170	0.173	95.15
3	37	1110	0.120	96.64
4	39	1170	0.146	95.91
5	35	1050	0.150	95.80

.

Trials 2–4 yield a mean freshwater flow rate of 1150 mL/h with a standard deviation of 30 mL/h, which is 4.2% below the AMESim simulation prediction of 1200 mL/h. This discrepancy is physically consistent with two simplifications inherent in the simulation model. First, the AMESim check valves are modelled as ideal elements that switch instantaneously between fully open and fully closed states, whereas in the physical prototype the ball-type check valves have a finite response time governed by spring stiffness and ball inertia; during the brief transitional period at each stroke reversal, partial backflow through the closing valve reduces the net displaced volume per cycle. Second, the 2/3 directional valve spool operates with a finite diametric clearance, whereas the simulation treats the valve as having zero internal leakage; in practice, a small cross-port leakage flow exists throughout the pumping cycle, further reducing the net seawater delivery to the membrane. Both effects are intrinsic to the physical construction of the device rather than operational variability, which is consistent with the low inter-trial standard deviation of 30 mL/h (coefficient of variation 2.6%) observed across Trials 2–4.

The standard deviation of 30 mL/h (2.6% of the mean) indicates good repeatability across Trials 2–4, confirming that the manual pumping cadence is consistent and that the device operates in a stable regime under these conditions.

Trials 2–4 yield a mean desalination rate of 95.90% with a standard deviation of 0.75%. The mean residual product salinity is 0.146%, corresponding to a total dissolved solids (TDS) concentration of approximately 1.46 g/L (expressed as NaCl). This value is well below the seawater feed concentration of 35.7 g/L, representing a salt rejection of about 95.9%, and lies below the WHO short-term acceptability threshold of 5 g/L TDS recommended for emergency drinking-water supply. It should be noted, however, that the corresponding sodium concentration (~0.57 g/L) still exceeds the WHO taste-based guideline of 200 mg/L and the 500 mg/L threshold; the product water is therefore intended as an emergency, short-duration potable source rather than as a long-term drinking-water supply. Further reduction in the residual salinity, for example by increasing the operating pressure or adopting a two-stage configuration, is identified as a direction for future improvement.

Two membrane-related phenomena that were not explicitly modelled in the present study but that influence the observed desalination performance deserve comment. First, concentration polarisation—the accumulation of rejected salt in a boundary layer adjacent to the membrane surface—locally elevates the wall salinity above the bulk feed value and therefore reduces the effective net driving pressure, contributing to the gap between the manufacturer-rated rejection (96–99.5%) and the measured steady-state value of 95.9%. This effect is expected to be more pronounced in the present device because the low, intermittent manual flow provides limited cross-flow to sweep the boundary layer. Second, membrane fouling—the progressive deposition of foulants on the membrane—would, over extended operation, further reduce permeate flux and rejection. In the intended emergency application, where continuous operation is measured in days to weeks rather than months to years, long-term fouling is not the governing constraint, and simple pre-settling or low-pressure pre-filtration of the raw seawater would be sufficient to protect the element over the required service interval. A dedicated long-term fouling and membrane-replacement study under real seawater conditions is nevertheless identified as important future work.

The relatively low DR in Trial 1 (90.14%) is a direct consequence of first-time membrane priming: the membrane housing contained a small volume of saline water from the manufacturing and assembly process, which mixed with the first batch of product water before the system reached a steady operating condition. This is a well-known start-up effect in spiral-wound RO systems and is not indicative of the membrane’s actual salt rejection capability. Under sustained operation, the RO membrane’s inherent salt rejection (specified at 96–99.5% for seawater-grade TFC polyamide membranes at 50 bar) is well above the observed steady-state value of 95.9%, suggesting that the membrane is operating within, but not at the upper bound of, its rated performance range. The gap between the manufacturer’s rated rejection and the measured value is attributable to the sub-optimal operating pressure (the circuit pressure was not independently measured but is expected to be somewhat below 50 bar due to the internal losses) and possible minor bypass through the imperfect seals of the prototype membrane canister.

Table 9. Comparison of simulation predictions, experimental results, and design targets.

Performance Metric	Simulation	Experiment (Mean, Trials 2–4)	Design Target
Freshwater flow rate (L/h)	1.20	1.15 ± 0.03	~1.20
Deviation from simulation	—	−4.2%	<10%
Desalination rate (%)	N/A	95.90 ± 0.75	>95%
Product water salinity (%)	N/A	0.146 ± 0.027	<0.18%
Overall dimensions (mm)	200 × 128 × 63	200 × 128 × 63	≤210 × 135 × 70
Energy saving ratio (%)	49.5	N/A	>40%

Note: N/A, not evaluated. Desalination rate and product water salinity were not predicted because membrane salt transport was not included in the AMESim model. The experimental energy-saving ratio was not measured because the prototype was not instrumented with a force sensor.

compares the key experimental results (Trials 2–4 mean) with the AMESim simulation predictions and the original design targets.

The agreement between simulation and experiment is satisfactory: the flow rate prediction is within 4.2% of the measured value, which is well within the acceptable engineering tolerance given the simplifications inherent in the AMESim model (ideal check valves, zero spool leakage, uniform seawater properties). The desalination rate exceeds the design target of 95% in all three valid trials. The device dimensions match the design exactly, confirming that the SolidWorks (version 2021; Dassault Systèmes SolidWorks Corp., Waltham, MA, USA) model translates faithfully into the physical prototype.

The single-pass recovery ratio of the device can be estimated directly from the pump geometry. Since the net permeate volume delivered per stroke equals the rod-swept volume while the seawater volume drawn in equals the blind-end swept volume, the geometric recovery ratio is r = d²/D² = (3.7/6)² ≈ 0.38. It should be emphasised, however, that this figure differs conceptually from the recovery ratio of a conventional RO plant, because in the present device the high-pressure reject brine is not discarded but is instead redirected into the rod-end cavity for energy recovery; the brine and permeate streams therefore remain coupled through the piston kinematics rather than being independently metered. The relatively small membrane area of the compact single element limits the absolute permeate throughput, and consequently the effective freshwater recovery per unit of feed processed remains modest. Increasing the membrane area—for example by adopting a higher-area module—would raise the achievable recovery, but would simultaneously increase the volumetric flow that must be pressurised per stroke and therefore the manual cranking effort. The chosen geometry (d/D ≈ 0.62) thus reflects a deliberate compromise between freshwater yield, energy-recovery ratio, and sustained hand-operability, as formalised in Equation (4).

As summarised in

Table 10. Comparison of the proposed device with representative portable and small-scale RO desalination systems.

Device	Actuation	Freshwater Output	Salt Rejection	Energy Recovery	Power Requirement	Portability/Scale
This work	Manual (hand-operated lever)	~1.15 L/h	>95%	Integrated hydraulic brine recovery (49.5% simulated)	None (fully off-grid)	Hand-held, 200 × 128 × 63 mm, ~450 g
Katadyn Survivor 06 [18]	Manual (hand pump)	~0.5 L/h	98.4%	None	None (fully off-grid)	Hand-held
Compact powered device [21]	Powered (onboard pump)	Not reported	Not reported	None	Onboard power required	Portable, self-contained
ROWPU-class system [19]	Motor-driven	High (industrial)	High	Yes (large-scale ERD)	Tens of kW	Truck/rail/airlift scale

, existing portable RO systems occupy two extremes. Hand-operated units such as the Katadyn Survivor 06 (Katadyn Products AG, Wallisellen, Switzerland) achieve excellent salt rejection (98.4%) but deliver only about half the freshwater output of the present device and incorporate no active energy recovery, so the entire pumping effort must be supplied by the operator against the full membrane back-pressure. Motor-driven ROWPU-class systems offer far higher throughput but require tens of kilowatts of generated power and vehicle-scale logistics, disqualifying them from individual carriage. Compact powered portable units integrate an onboard pump within a self-contained housing but rely on a stored power source and likewise incorporate no brine energy-recovery mechanism. Against this background, the distinguishing feature of the present device is that it is the only hand-held, fully manual unit among those surveyed to integrate a passive hydraulic brine energy-recovery circuit. The 49.5% brine-assisted force saving demonstrated in simulation represents a meaningful ergonomic advantage: for sustained manual operation in emergency maritime conditions, approximately halving the pumping effort can substantially extend the operational endurance of an individual user, while the achieved output of ~1.15 L/h at >95% rejection is sufficient to meet individual short-term drinking-water needs.

4. Conclusions

This study presented the design, AMESim hydraulic simulation, and experimental validation of a compact, manually operated seawater desalination device integrating a brine hydraulic energy recovery mechanism. The following principal conclusions are drawn:

(1)

A valve-commutated piston pump coupled with a mechanically actuated 2/3 directional valve successfully implements an integrated brine energy recovery function without any electrical components. The lever handle mechanism, with a 5.7:1 mechanical advantage confirmed by Grübler’s criterion to possess one degree of freedom, reduces the effective grip force at the handle tip to approximately 25 N and 13 N during the intake and delivery strokes, respectively.

(2)

AMESim simulation predicted a steady-state freshwater output of 1.20 L/h. The brine-assisted energy recovery reduces the delivery-stroke RMS piston force from 99 N to 50 N, achieving a cycle-averaged energy saving of 49.5%. The 2/3 directional valve correctly commutated the brine circuit in both stroke directions with no measurable steady-state flow loss, as confirmed by the port flow analysis.

(3)

Prototype experiments with 3.57% artificial seawater at a pumping cadence of 1 Hz produced a mean freshwater flow rate of 1150 mL/h and a mean desalination rate of 95.9%, in close agreement with simulation predictions. The product water salinity of 0.146% (≈1.46 g/L TDS) lies below the WHO short-term threshold for emergency drinking-water supply, confirming the device’s suitability for short-term individual emergency use, although the residual sodium level indicates it is not intended as a long-term supply.

Future work will focus on long-term membrane fouling characterization under real seawater conditions, experimental verification of the brine-assisted force reduction through direct force measurement on the prototype, and further optimization of the lever geometry and valve switching timing to improve both ergonomic performance and freshwater yield.