Automated multiport flow-through water pumping and sampling system.

Flow-through systems are often used in aquarium and aquaculture facilities, laboratories, and aboard research vessels and other mobile systems to collect, analyze, and monitor water properties as they vary across time and location. These systems most often intake water from a single source and deliver it to a suite of flow-through sensors after which waste water either exits the system or is recirculated back to the source. Here we describe a system that is designed to take water from multiple sources via a multiport valve manifold and deliver it to a common sample stream, facilitating analysis by a single suite of flow-through and probe type sensors. Build cost depends on the manifold design and the number of valves, but generally under $9000. The inclusion of a Free Surface Interface Cup (FSIC) allows probe type sensors or sample "sippers" that require unpressurized conditions to be utilized down-stream of the pumping system and manifold. With the exception of the multiport sampling manifold, all components of this system are available off-the-shelf, simplifying construction, service, and maintenance. The operating system code is open source and based on the Arduino platform, enabling users to customize the code to better fit their requirements.

Hardware in context

One of the more commonly used field methods for continuous water sampling utilizes shipboard underway pumping systems. Research vessel operators have installed flow-through sensors (available from various manufacturers and/or custom built) into their underway sea water systems for many years to monitor surface seawater properties during ship transits [1]. Typically utilizing a single fixed intake (i.e., at a fixed level on the hull), these systems can be used for continuous observations for many weeks and over great distances while continuously sampling, allowing scientists to better understand ocean surface variability. The combination of underway and discrete station sampling at depth can provides complementary multidimensional perspectives on oceanographic parameters and processes.

Research facilities also use seawater intake systems to feed tanks and aquaria where scientists attach instruments and sensors so they can monitor their system water supply. Some systems are designed for sampling at a single location with the inlet water stream delivered to a suite of sensors. Many fixed location observing systems (buoys, benthic tripods, and fixed moorings) would benefit from the addition of a reliable multi-stream sampling capability.

The inspiration for this design was taken from Genin et al. [2] who used a series of 10 pumps on each of 4 moorings to deliver water to shore for discrete chlorophyll measurements. This system was not automated and used 40 surface powered submerged water pumps and substantial lengths of tubing. To maintain flow, the submerged pumps required periodic biofoul cleaning. In 2002 we evaluated this system in the field and while functional, it was labor intensive and inefficient for many other analyses. We developed and built an automated version of the Genin et al. [2] system that eliminates the use of many, separate submersible pumps by using a multiport manifold approach as reported in Teneva et al. [3]. This automated sampling system utilized in Palau was comprised of 24 water sampling tubes (6 tubes at different depths on each of 4 moorings) and performed high resolution real-time carbon system measurements. The Automated Multiport Pumping System (AMPS) simplifies the water sampling and analysis by reducing the number of pumps required, providing ease of deployment and maintenance, reducing biofouling potential, and facilitating real-time and rapid high resolution data acquisition.

The advantage of the AMPS is that only one set of calibrated sensors/instruments are required to analyze multiple parameters, reducing the need to use or to cross calibrate duplicate instrumentation, and thereby improve data quality. In Fig. 1 the AMPS pumps water through a thermosalinograph (TSG) a pCO2 sensor and then into a Free Surface Interface Cup (FSIC) housing a dissolved oxygen optode, pH probe, and a sipper tube delivering water samples to a dissolved inorganic carbon analyzer. Other flow-through sensors could be used to measure properties such as, fluorescence, turbidity, chlorophyll-a, oxidation-reduction potential, chloride, algae, ammonium, and nutrients. The FSIC provides the means to add other probe type sensors and sipper tubes for other water quality measurements.

Fig. 1

Example setup of the AMPS (A) and a schematic (B) showing the various components and the water pathway (blue lines with arrows) though the system. The principle AMPS components includes a valve controller (1), multiport sample manifold (2, SM) with DAVs (V), variable speed controller (VSC) with peristaltic sampling pump (3), and waste pump (4) deployed on a 22 foot vessel in 2018, powered by a Honda 2000 W generator. In this example, the AMPS is delivering seawater from 8 depths to a Sea Bird Electronics SBE45 TSG (5), Pro Oceanus Pro-pCO2 analyzer (6), FSIC (7) containing a Honeywell Durafet pH probe (8a & 8b) an Aanderaa 3850 dissolved oxygen optode (9), a custom Stanford University designed dissolved inorganic carbon (TDIC) analyzer (10), and a digital flow meter (FM). The * denotes optional instrumentation used to test this system during development and for field research. A SCUBA tank with 2-stage regulator provided compressed air to power the pneumatic valves (90 psi) and to an in-line single stage regulator that was to step down the carrier gas flow (10 psi) for the TDIC analyzer. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Our main interest in designing the AMPS was to perform high resolution real-time carbon system chemistry measurements where the principle components of the AMPS, such as the valve controller, pumping system and multiport manifold, can be used for other purposes. For example, the same pumping flow rate may not be required for some users. We therefore describe two different sampling manifold designs, with ¼” and ½” valves for low and high flow rate applications, respectively. Reducing the number of diverter air valves (DAV) on the manifold and/or using the smaller valves will reduce cost and any unused ports can easily be plugged. A lower flow sampling and waste pump can be selected in place of the units we used to further reduce the cost of the system as long as they remain positive displacement pumps capable of maintaining prime and a consistent flow rate. The Arduino Mega 2560 R3 controller has numerous general-purpose input/output (GPIO) pins allowing for system expansion, including the addition of more valves, Wi-Fi, Ethernet, flow meters, and other sensors. If more displayed information is required, a larger LCD can be attached to the controller or a computer can be connected for real-time monitoring. The AMPS design we present here should provides a flexible platform that can be customized to serve many applications.

The AMPS featured in Fig. 1 is an electronically-evolved version of the system used in Teneva et al. [3]. It incorporates easily acquired components and can also operate autonomously. In practice, the AMPS has been used in various configurations, deployment settings, and data obtained from this system have been published in several journals [3], [4], [5], [6] making it a well-tested system.

Hardware description

The AMPS consists of a valve controller, multiport sample manifold, pumping system, and FSIC. With the exception of the custom built inner multiport valve manifold, all of the other components are available for off-the-shelf purchase and are substitutable depending on the needs of the system or user. We maximized the number of ports used for our research projects, but fewer valves could be used, thereby lowering the total cost of the system, while retaining capacity for expansion.

A. Valve controller: (Fig. 2): The valve controller is housed in a corrosion resistant fiberglass waterproof enclosure with fiberglass back plate used to mount the electronics along with watertight electrical connectors, stainless steel hardware, and components for protection from the environment. Inside the valve controller enclosure is a Clippard 12-channel electronic valve assembly with solenoids and booster solenoids. The booster solenoids provide sufficient operating air pressure pulse to “lock” the valve pistons into place. Between 75 and 90psi (5.2 to 6.2 bar) air pressure is supplied for this purpose from a 2-stage regulator attached to a high-pressure air source. Since the AMPS is pulling water through the assembly instead of pushing, suction on the pistons against the springs in the DAV can be overcome, allowing three ports in a DAV to be open at the same time. The addition of the booster valves to the solenoid valves fixes this problem. An air compressor, building air, or high-pressure tanks (including SCUBA tanks) can be used with 2 stage regulators to run the valves. The benefit of the Clippard 12-channel electronic valve assembly is that it has all of the needed circuitry on the board, is compact, and affordable.

Fig. 2

Valve controller in waterproof enclosure equipped with Arduino 2560 R3 Micro controller with 16x02 LCD screen (1), RTC (2), and Micro SD card reader (3), 16-channel relay board (4), 12-channel electronic valve assembly board, and the 12VDC (6) and 24VDC (7) power supplies.

Electronics – The Clippard 12-channel electronics and pneumatic valve assembly is controlled by an Arduino Mega 2560 R3 controller with a 16x02 LCD screen, a real-time clock (RTC), Micro SD card reader, and a Sainsmart 16-channel relay board. A DIN-Rail-mounted 24VDC, 40 W, 1.7A power supply supports the electronics of the Clippard 12-channel valve assembly and the bus on the 16-channel relay board. A second DIN-Rail-mounted 12VDC, 20 W, 1.7A power supply supports the Arduino 2560 R3, RTC, Micro SD card reader, LCD screen, and logic portion of the 16-channel relay board. Separate DIN-Rail mounted power supplies are compact and failed units are easy to replace. The 16 character 2-line LCD screen allows the system to be monitored at a glance while in autonomous or real-time mode. The RTC is necessary for proper timing and the Micro SD card stores all recorded actions. The Arduino hardware and software is simple, open source, and inexpensive. Options such as Wi-Fi and Ethernet can be added to the Arduino Mega 2560 R3 but are not discussed here.

B. Pneumatic Valve Manifold (Fig. 3): Pneumatic valves allow for multiple valve configurations while minimizing exposure of the electronics to the elements. We use the RK Industries ½” NPT 3-way DAVs for most of our builds, but have also used the smaller ¼” 3-way DAV. The DAVs have a polyvinyl chloride (PVC) body and piston, are equipped with Viton o-rings, stainless steel springs, making them corrosion resistant. All of the pneumatic valves are supported and connected with plastic fittings and connectors. We opted for plastic connections and tubing to eliminate corrosion issues and for improved water quality.

Fig. 3

Multiport valve sampling manifold with the ½” DAV design on the left and the ¼” DAV design on the right.

Inner multiport sampling manifold – The only custom-built component in this design is the two-piece Delrin inner multiport sampling manifold body. We chose a circular manifold design over a less expensive “stick” manifold design because the circular design is more compact and the two-piece design makes it easier to clean. The outside dimensions of the manifold are 6.00″ (15.24 cm) in diameter, 2.75″ (6.985 cm) in height. Inside dimensions are 4.50″ (11.43 cm) in diameter by 1.25″ (3.17 cm) in height yielding an internal volume of 325 ml. The valve input, top, and drain ports on the manifold are sized for the valves used (i.e., ½″ NPT for the ½″ DAV and ¼″ NPT for the ¼″ DAV).

Outer valve waste manifold – The waste manifold is circular to match the inner manifold and is plumbed so that wastewater is controlled by a single positive displacement impeller pump (Fig. 3). The waste manifold is connected to the normally open port on the DAVs using Nylon compression tube fittings along with PVC nipple and plug fittings. With the exception of the DAV input ports which are sized for their respective NPT size, the outer manifold components are similar on both the ¼″ and ½″ designs. To facilitate purging of the waste lines, the waste manifold is bisected as illustrated in the 8-port example in Fig. 3B where an elbow is in place at valves 4 and 5. Completely assembled, the overall diameter of the ¼″ and ½″ design are identical at 20″ (50.8 cm). The unconnected Tee fitting on the waste manifold (Fig. 3B) connects to the waste pump priming manifold as shown in Fig. 4.

Fig. 4

Waste pump priming manifold schematic. Priming pump (A), back filling input sample tubes and DAVs on multiport sampling manifold (B), pulling water from input sample tubes to waste (C), final configuration pulling water from all sample tubes to waste and delivering water to sensors and instruments using peristaltic sampling pump (D). The blue lines and arrows indicate the water flow. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

C. Waste pump and priming manifold (Fig. 4): The AMPS uses a positive displacement impeller waste pump that moves water at the same rate regardless of the pressure on the inlet and it can create a vacuum on the inlet side allowing for suction lift unlike a centrifugal pump. The impeller pump we selected is self priming up to 2 m. The waste pump has a factory flow rate specification of ~ 22 L min−1 and when connected to all 12 ports the theoretical exit flow rate is ~1.83 L min−1 for each tube. This is obtained by dividing the pump flow by the number of ports. It is important to measure the actual wastewater output flow per port with all tubes connected as this will differ from the theoretical pump specifications as performance is influenced by the length and diameter of tubing utilized. Connected to the waste manifold is a priming manifold that uses two manual 3-way ball-type diverter valves to facilitate priming the sample lines connected to the DAVs as shown in Fig. 4.

D. Sampling pump (Fig. 5): The AMPS uses a variable speed high volume peristaltic pump that is self-priming and can be adjusted to match the flow rate of any manifold configuration. No metal components are in contact with water. The variable speed controller for the peristaltic pump is not waterproof. To protect it from the elements we opted to install it inside a corrosion resistant fiberglass waterproof enclosure. For proper synchronization the sample lines connected to each DAV need to be of equal length and the peristaltic sampling pump flow rate needs to be adjusted to match the outflow of the waste pump divided by the number of DAVs.

Fig. 5

Peristaltic pump (1) used for water sampling. The variable speed controller (2) was installed in a waterproof fiberglass enclosure (3) because the electronic controller is not waterproof.

E. Free Surface Interface Cup (FSIC): The FSIC (Fig. 6) is essentially a cup in a funnel and is fabricated from simple PVC and acrylonitrile butadiene styrene (ABS) plumbing materials. The cup volume can be adjusted to accommodate different sampling and sensor needs. The FSIC provides access to an unpressurized water stream, required for some analytical instruments. Water coming from the peristaltic sampling pump enters the bottom of the inner cup, overflows and drains out the bottom of the outer funnel. When using flow-through sensors with the FSIC they should be placed upstream of the FSIC which is always plumbed downstream of the peristaltic pump. The FSIC can be eliminated if not needed.

Fig. 6

FSIC is a cup in a funnel design used for sampling unpressurized water exiting the enclosed portion of the flow-through system. The FSIC is fabricated from PVC and ABS components.

F. Flow meter: We use a Proteus Industries Inc. FluidVision USB 6000 series flow meter that can log data to a computer. It can measure flow rates of 0.4 to 4 L min−1 and requires a dedicated computer. As of this submission, we have not configured the Arduino Mega 2560 R3 and software to log flow rates in autonomous mode, but such logging is possible using the extra GPIO ports on the controller.

Design and software files

Design and software files for the AMPS are available for download from the Open Science Framework. The file AMPS Drawings and Schematics.pdf (https://doi.org/10.17605/OSF.IO/CZJ5K) contains all of the drawings and schematic diagrams to build the AMPS. The AMPS software was coded using the Arduino IDE platform that is available via download from https://www.arduino.cc/en/Main/Software and links to download the AMPS code is provided in the table below.

AMPS Drawings and Schematics – contains all of the drawings and schematic diagrams to build the AMPS.

time_set_manually – Used to set the real-time clock.

current_time_LCD – Used to display the time from the real-time clock on computer and LCD.

AMPS-V2_main_RTC_LCD – Used to run the valve controller after the real-time clock is set.

Bill of materials

A bill of materials list for the valve controller, multiport sampling manifold, waste pump priming manifold, peristaltic pump controller, and FSIC are available for download from the Open Science Framework in pdf format under filename AMPS Build of Materials.pdf (https://doi.org/10.17605/OSF.IO/CZJ5K). Most of the items in the bill of materials list can be acquired from alternate suppliers as needed or preferred. In some instances, it was less expensive to buy components in bulk (e.g., stainless steel fasteners, PVC pipe) than the exact number of items from a local hardware store.

Build instructions

A separate Build Instruction document is available for download from the Open Science Framework under filename AMPS Build Instruction Manual.pdf (https://doi.org/10.17605/OSF.IO/CZJ5K). This manual includes the step-by-step assembly and wiring instructions for all components and the tools needed to assemble the system.

Operation instructions

The separate User Manual for the AMPS is available for download from the Open Science Framework under filename AMPS Users Manual.pdf (https://doi.org/10.17605/OSF.IO/CZJ5K). This manual provides the step-by-step operations for the entire system.

Validation and characterization

The AMPS has been deployed in the Republic of Palau, Palmyra Atoll, American Samoa, a large aquarium tank at Hopkins Marine Station, and as a research vessel lab-mounted underway system in Antarctica. Published results are available [3], [4], [5], [6], [7] making this a proven and well-tested system. The system has seen continuous use, for up to 8 weeks in Antarctica (2018) and in Monterey, California (2018) with results in review [8]. To demonstrate the utility of this system, we show a portion of the Palmyra Atoll 2012 field season data [4]. In this experiment, the AMPS was deployed on a 3 m × 4 m raft anchored in a backreef environment protected from the elements by a 3 m × 3 m gable-roof tent (Fig. 7). This experiment was conducted at 6 m water depth with a tube placed 3 m above the bottom on each of 4 moorings. For the purpose of this example Tubes 2 and 4 will be used as both mooring were instrumented with ability to measure both temperature and salinity.

Fig. 7

AMPS deployed on a raft on a backreef at Palmyra Atoll. The system was powered with a 5KW generator and compressed air provided by a SCUBA tank. Two of the four moorings are visible in the foreground.

The AMPS was setup as shown in Fig. 7 and Fig. 1B illustrates the instrumentation used in this experiment. For research quality data acquisition it is imperative that the water measured via the AMPS reflects in-situ water properties. A small portion of the Palmyra 2012 Tube 2 and 4 temperature and salinity data collected from the TSG is compared to matching time data from in-situ thermistors (SBE-56, Sea Bird Electronics) and temperature, salinity, pressure sensors (SBE-37-SMP, Sea Bird Electronics) as an example, and is displayed in Fig. 8. The salinity obtained from the SBE-37-SMP (black line) and TSG (red line) show strong agreement (<0.02 overall offset) for both Tubes 2 and 4 (Fig. 8). The offset is attributed to a slight calibration difference between the TSG and the SBE-37-SMP. The SBE-56 temperature measured by the top (Orange line) and bottom (blue line) are very similar and shows an offset of ~0.1 °C during the daytime hours and ~0.04 °C at night for Tubes 2 and 4 on the TSG (green line, Fig. 8).

Fig. 8

SBE-56 temperature (blue and orange) and SBE-37-SMP salinity (black) comparison with the on-deck SBE-45 TSG temperature (green) and salinity (red) data from the AMPS. The overall salinity offset between the SBE-37SMP and the SBE-45 TSG is <0.02 PSU. The offset for the daytime temperature readings between the SBE-56 thermistors and the SBE-45 TSG is ~0.09 °C during the day and ~0.04 °C at night. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The data presented in Fig. 8 extends from 0900 local time on 09/21/12 to 1730 local time on 09/22/12. Weather conditions at the site during the experiment were overcast with two squall events. The impacts of both squall events are shown as negative excursions in salinity and temperature in Fig. 8. Note the delayed response between the water temperature and salinity. The first squall was shorter and less intense than the second squall which lasted for several hours and delivered cooler air and heavy rain. As the weather cleared the TSG readings were closer to the in-situ probe measurements. Cooler rainwater impacted the water in the tubes that were floating on the surface, contributing to an offset between the TSG and the in-situ instruments. To minimize this artifact, sampling tubes should be deployed at depth and then rise through the interface to the manifold to limit tube exposure to ambient conditions.

Summary

Accessibility – There are many applications for an automated flow-through pumping system like the AMPS. Our research group has used this system and its variants for multiple field projects (Table 1) in several different environments (coral reef, kelp forest, tank, cold water, and ice-covered seas) and platforms (beach, raft, deck of a small vessel, and a lab on a research vessel). This system has been extensively tested in the field to demonstrate its versatility and portability. The cost of the pumping system is reasonable, estimated to be less than $9000 without the TSG and other sensors. If fewer DAVs are needed, costs can be reduced by ~$300/DAV and even less if smaller pumps are used.

Table 1

Published and exploratory work using the AMPS in various configurations including year, location, and equipment used for each experiment.

Field Season	Location	Environment	Platform	Equipment used	Citation
2011	Koror, Palau	Back reef	15 m Sport fishing vessel	Dual TDIC, pH, TSG, FSIC, Manifold	[3]
2011	Palmyra Atoll	Back reef	Raft	Single TDIC, pH, DO, TSG, FSIC, Manifold	Exploratory work
2011	Ofu, American Samoa	Back reef	Beach	pH, DO, TSG, FSIC	[5]
2012	Palmyra Atoll	Back reef	Raft	Single TDIC, pH, DO, TSG, pCO2, FSIC, Manifold	[4]
2013	Ross Sea, Antarctica	Polynya	R/V Palmer - underway system	Single TDIC, TA, DO, pCO2, FSIC	[7]
2014	Patagonia, Chile	Fjords, Lago Sarmiento	R/V Neecho – 12 m vessel	Single TSG, FSIC, pH	Exploratory work
2015	Koror, Palau	Sea grass	4 m vessel	Single TSG, FSIC, pH	Exploratory work
2015	Koror, Palau	Tank	Outside tank facility	Single TSG, FSIC, pH, TA, TDIC	Exploratory work
2015	Hopkins Marine Station, Pacific Grove, CA	Tank	Outside tank facility	Single TDIC,pH, DO, TSG, pCO2, FSIC	[6]
2018	Hopkins Marine Station, Pacific Grove, CA	Kelp Forest	4 m and 7 m support vessel	Single TDIC,pH, DO, TSG, pCO2, Manifold, FSIC	[8]
2018	Ross Sea, Antarctica	Polynya	R/V Palmer – underway system	Single TA, DO, FSIC	Exploratory work

Limitations: We have used this system in the warm backreef coral setting of Ofu, American Samoa (30+°C), on the coast of Monterey, CA (8 °C to 10 °C), in Patagonia lakes and fjords (3 °C to 5 °C) and Antarctica (−2 °C to 0 °C) with no issues with water in the lines or with the hardware. The main concerns are with the ambient conditions. As seen in the Palmyra 2012 experiment example (Fig. 8), cooler rain caused a slight difference in TSG reading compared to in-situ instrumentation. Deploying the tubes at depth will minimize this impact provided that in doing so this will not damage the environment (e.g., break corals). Extreme cold ambient conditions may also be a limiting factor where freezing tubes or ice in the FSIC could be an issue. Heat can also be a concern where thermal shock to non climate controlled instruments is a possibility and if sample tubes are deployed on a hard surface they can get hot (e.g., sandy beach, dock, or vessel deck) and water properties may be altered.

Strong currents can make deployment of tubing complicated and if there is high water turbulence, bubbles can impact flow-through sensors and probes as well as some analytical instrumentation. Although we have not deployed a debubbler (e.g., MSRC VDB-1 Vortex Debubbler available from SoMAS, Stony Brook University, Stony Brook, NY) in our system, one could be installed to minimize bubbles in the analytical water stream. Since we analyze TDIC and pCO2 we avoid using a debubbler and take additional steps to minimize bubble formation. High rates of primary production can produce oxygen bubbles that can affect sensors and sample intake lines placed inside the FSIC. To minimize this impact, we have used a 5 μm to 100 μm Sumpsocks to minimize bubble density.

Final Assessment: The AMPS is portable, compact, versatile, simple, and dynamic, allowing it to be used in many settings and easily tailored to meet researcher’s needs. The design is straightforward allowing fabrication by those with minimal mechanical and electrical knowledge. With the exception of the inner sample manifold, the components used are available for purchase off-the-shelf, affordable, and easily available. We feel confident that the pumping system we describe here will provide and deliver water for research quality analyses. The results from past research projects also demonstrates the versatility of this design without analytical compromise.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Specifications table

Hardware name	Automated Multiport Pumping System (AMPS)
Subject area	• Oceanography and Limnology • Chemistry and Biochemistry • Biological Sciences • Environmental and Aquaculture Sciences
Hardware type	• Measuring water physical and chemical properties • Field and laboratory measurements and sensors
Open Source License	GNU General Public License (GPL) 3.0
Cost of Hardware	$8736 (1/2″ valve system) or $8455 (1/4″ valve system)
Source File Repository	https://doi.org/10.17605/OSF.IO/CZJ5K

Design file name	File type	Open source license	Location of the file
AMPS Drawings and Schematics	pdf	GNU General Public License (GPL) 3.0	https://doi.org/10.17605/OSF.IO/CZJ5K
AMPS Build of Materials	pdf	GNU General Public License (GPL) 3.0	https://doi.org/10.17605/OSF.IO/CZJ5K
AMPS Build Instruction Manual	pdf	GNU General Public License (GPL) 3.0	https://doi.org/10.17605/OSF.IO/CZJ5K
AMPS User Materials	pdf	GNU General Public License (GPL) 3.0	https://doi.org/10.17605/OSF.IO/CZJ5K
time_set_manually	ino	GNU General Public License (GPL) 3.0	https://doi.org/10.17605/OSF.IO/CZJ5K
current_time_LCD	ino	GNU General Public License (GPL) 3.0	https://doi.org/10.17605/OSF.IO/CZJ5K
AMPS-V2_main_RTC_LCD	ino	GNU General Public License (GPL) 3.0	https://doi.org/10.17605/OSF.IO/CZJ5K