{"id":10840,"date":"2026-11-06T08:32:13","date_gmt":"2026-11-06T08:32:13","guid":{"rendered":"https:\/\/www.herewinpower.com\/blog\/2026-10-27-underwater-robot-battery-auv-rov-selection\/"},"modified":"2026-11-06T08:32:13","modified_gmt":"2026-11-06T08:32:13","slug":"2026-10-27-underwater-robot-battery-auv-rov-selection","status":"publish","type":"post","link":"https:\/\/www.herewinpower.com\/nl\/blog\/2026-10-27-underwater-robot-battery-auv-rov-selection\/","title":{"rendered":"How to Choose a Battery for Underwater Robots: AUV vs. ROV"},"content":{"rendered":"
\"Illustration<\/figure>\n\n\n\n

An underwater robot battery must deliver reliable power under hydrostatic pressure, in seawater that can compromise insulation and sealing. Pressure rises by roughly one atmosphere for every 10 m of seawater (NOAA PMEL<\/a>), so a 300 m vehicle and a 3,000 m vehicle load the same enclosure very differently.<\/p>\n\n\n\n

Cell datasheets can’t answer that question on their own, because the BMS, wiring, connectors, sealing and pressure-protection structure all add mass and volume the vehicle has to carry. Depth ratings and pack-level verification therefore belong in the specification from the start. Platform matters too: an AUV carries its energy throughout the mission, while an ROV may draw power down an umbilical and use its battery only for backup or peak support. Two rules follow, and both recur throughout this guide \u2014 an IP rating is not a depth rating, and cell data is not pack performance.<\/p>\n\n\n\n

Define the Power Requirements of the Underwater Robot<\/h2>\n\n\n\n

Before comparing cells, establish what the battery is being asked to do on this specific vehicle. Battery sizing begins with two considerations: the vehicle’s mission profile and its power architecture. The mission profile defines the required endurance, operating depth and load pattern, while the power architecture determines how much of the vehicle’s power demand the battery must supply. If either is overlooked, the pack may meet its rated specifications but still fall short of the vehicle’s actual operating requirements.<\/p>\n\n\n\n

AUVs: Balancing Energy, Endurance, and Mass<\/h3>\n\n\n\n

An AUV carries its own energy, so endurance is a mass-and-volume problem as much as an energy problem: every kilogram of pack has to be offset by displacement, and every litre takes space from payload or lengthens the hull. The number to size against is usable energy at the end of the mission, reserve included, not nominal capacity. Capacity and deliverable power are separate questions: a pack can pass a capacity test and still fail under load.<\/p>\n\n\n\n

A power budget usually covers four load groups: propulsion; sensors and payload such as sonar, cameras and sampling systems; navigation and control electronics; and auxiliary loads such as lighting, data logging and thermal management. Thruster draw shifts between transit, hovering and maneuvering, while some electronics run continuously. Estimate mission duration, average and peak power per group, and the reserve you want at recovery. That shows whether extra capacity actually buys endurance once pack mass is counted, and whether the pack still fits the hull and trim budget.<\/p>\n\n\n\n

MBARI’s Tethys long-range AUV illustrates how far the range stretches: a 3.8 kWh rechargeable pack for missions of about a week or 600 km, and a 12 kWh primary configuration for about three weeks or 1,800 km (MBARI Tethys-Class Long Range AUV paper<\/a>). Mission duration, not cell choice, sets that budget first.<\/p>\n\n\n\n

ROVs: Identifying the Battery’s Role in the Power Architecture<\/h3>\n\n\n\n

An ROV is tethered, and the umbilical usually carries power from the surface, so the battery is rarely the primary source. In a purely umbilical-powered vehicle it may be absent or limited to a small backup for safe recovery. In a hybrid architecture it supplies peak power during thruster transients while the umbilical carries average load, which shifts the pack’s priority from energy to discharge capability. In a battery-backed design it may carry hotel loads through an interruption.<\/p>\n\n\n\n

Backup packs are easy to under-specify, because they don’t serve normal operating power. Size them around the emergency load profile instead: which functions must survive a tether failure \u2014 control electronics, navigation, lighting, controlled ascent or safe shutdown \u2014 and for how long. That sequence fixes the usable energy reserve, the peak discharge capability and the minimum voltage the pack must hold during recovery.<\/p>\n\n\n\n

A peak-shaving pack needs enough continuous and peak discharge capability, acceptable voltage stability under transients, and protection coordinated with the surface supply. Whichever role applies, the pack sits charged for long stretches, so standby behaviour belongs in the specification. Long-term storage at high state of charge accelerates degradation in some lithium-ion chemistries, so the storage SOC and maintenance interval should follow the cell manufacturer’s recommendations and be agreed with the battery manufacturer as part of the specification.<\/p>\n\n\n\n

Smaller platforms repeat the logic at a different scale. Some long-endurance gliders use primary lithium systems for their low self-discharge and long storage life, while inspection-class vehicles and consumer underwater drones put more weight on cost, packaging and sealing.<\/p>\n\n\n\n

Compare Battery Options at the Pack Level<\/h2>\n\n\n\n

With the requirement defined, the comparison moves from cells to the assembled pack. A common source of selection errors is comparing cell-level energy density without accounting for the additional mass and volume of the complete pack.<\/p>\n\n\n\n

Cell-Level vs. Pack-Level Energy Density<\/h3>\n\n\n\n

Gravimetric energy density (Wh\/kg) sets the mass cost of a given energy; volumetric energy density (Wh\/L) sets the space it occupies, and the two rarely rank options the same way. Neither is meaningful at cell level for an underwater vehicle, because a complete pack also includes the BMS, structural support, wiring, connectors, enclosure or pressure-compensation hardware, and any thermal management or buoyancy components. The cell-level vs. pack-level energy density comparison<\/a> covers which figure belongs in a specification.<\/p>\n\n\n\n

Pack-level numbers generally land below cell-level ones, and the gap varies with enclosure design, protection requirements, cell format and integration constraints. Calculate it from the proposed configuration rather than applying a universal percentage.<\/p>\n\n\n\n

That is also why “system mass per kWh” and “system displacement per kWh” are better concept-stage design inputs than a target pack energy density, since buoyancy, trim and layout can then be sized around the real pack.<\/p>\n\n\n\n

Matching Battery Chemistry and Cell Format to the Application<\/h3>\n\n\n\n

Chemistry follows from the vehicle’s energy, power, mass, volume and service-life requirements. Nickel-rich chemistries such as NMC and NCA suit projects where pack mass or volume is the binding constraint, but they carry thermal management, cycle-life and cell-specific safety requirements that have to be designed for. LFP trades energy density for cycle life and thermal stability, which fits high-cycle inspection and teaching platforms; its low-temperature behaviour depends on cell design, water temperature, discharge rate and thermal management, so ask for data at your operating temperature rather than disqualifying it in advance \u2014 the same rule that applies when comparing LiPo, Li-ion and LiFePO4 robotics packs<\/a>. Primary lithium cells serve long-duration single-use missions such as gliders, but they aren’t rechargeable.<\/p>\n\n\n\n

Format is a separate decision. Pouch, cylindrical and prismatic cells differ in mechanical support, packaging efficiency, thermal behaviour and integration constraints, and a pouch and a cylindrical cell of the same chemistry are not interchangeable in an underwater pack. Choose the format for integration and pressure architecture, not as a proxy for chemistry.<\/p>\n\n\n\n

\n\n<\/colgroup>

Candidate option<\/p><\/th>

Potential advantage<\/p><\/th>

Key trade-offs<\/p><\/th><\/tr>

NMC\/NCA<\/p><\/td>

Higher potential gravimetric or volumetric energy density<\/p><\/td>

Thermal management, cycle life and cell-specific safety requirements<\/p><\/td><\/tr>

LFP<\/p><\/td>

Cycle life and thermal stability advantages<\/p><\/td>

Energy density, size and low-temperature performance under specific conditions<\/p><\/td><\/tr>

Primary lithium<\/p><\/td>

Long-term energy storage potential for specific missions<\/p><\/td>

Single use, discharge characteristics and system constraints<\/p><\/td><\/tr><\/tbody>\n<\/table>\n<\/figure>\n\n\n\n

This table is a first-pass screen, not a recommendation. What fits depends on the vehicle’s requirements and on the pressure-protection architecture.<\/p>\n\n\n\n

Choose the Right Pressure-Protection Architecture<\/h2>\n\n\n\n

Pressure protection \u2014 how the pack is isolated from the surrounding water and hydrostatic pressure \u2014 also drives pack mass, available volume, maintenance and cost, which is why it belongs in the same discussion as chemistry.<\/p>\n\n\n\n

Pressure-Resistant vs. Pressure-Compensated Designs<\/h3>\n\n\n\n

A pressure-resistant enclosure keeps the pack near one atmosphere inside a rigid hull. The cells get a stable, dry environment, but wall thickness and mass grow with depth, and sealing, penetrators and pressure cycling become design concerns. The mass is easy to underestimate: extra buoyancy to offset a heavier hull takes up volume of its own, so hull and compensation volume belong in one block of the mass and layout budget.<\/p>\n\n\n\n

A pressure-compensated design lets external pressure act through a compensating medium, cutting the pressure difference across the enclosure. Usually an insulating fluid fills the enclosure while a flexible membrane or piston absorbs volume change as the pack descends and surfaces. That removes the need for a heavy hull but adds its own requirements: the fluid must stay compatible with cells, potting compounds and seals over the service life and hold its dielectric properties, the seal interface has to keep seawater out of the fluid path while letting the compensator respond, and the compensator sees fatigue loading on every dive. A review of pressure-compensated immersion batteries<\/a> reaches the same conclusion \u2014 the heavy cabin goes, but a corrosion-resistant shell, a durable compensator and clean insulating fluid remain essential.<\/p>\n\n\n\n

Neither architecture wins in general. The choice turns on four factors: operating depth and pressure-cycle requirements; the mass, volume and buoyancy budget; maintenance needs and material compatibility; and manufacturing and lifecycle cost. A shallow inspection vehicle serviced frequently may favour a different approach from a deep-water AUV built for long deployments.<\/p>\n\n\n\n

Seal failures rarely happen on the first dive. They appear after tens or hundreds of pressure cycles, as O-rings take a permanent set, the enclosure fatigues and penetrator threads loosen. A figure that states only a maximum depth therefore describes very little, so specify rated depth, qualification depth, and the cycle count and hold duration permitted at that depth.<\/p>\n\n\n\n

Pressure Ratings vs. IP Ratings<\/h3>\n\n\n\n

Maximum dive depth and an IP rating describe different things. An IP rating addresses water ingress under the test conditions written into that standard, not sustained hydrostatic pressure. The second digit in an IP67 rating covers temporary immersion as defined by the standard: under IEC 60529, the IPX7 test places an enclosure a metre below the surface for 30 minutes (IEC 60529 ingress protection testing<\/a>), which says nothing about repeated diving. Use the rating as an ingress-screening signal only. The target-depth pressure rating and pressure-cycling evidence are separate items to ask for.<\/p>\n\n\n\n

Verify Power Delivery, Temperature, and Electrical Safety<\/h2>\n\n\n\n

A pack that has enough energy can still fail the mission if it cannot deliver current without excessive voltage sag, if temperature limits cut usable capacity, or if the interface between pack and vehicle fails. All three are power-delivery problems rather than energy-storage problems.<\/p>\n\n\n\n

Continuous and Peak Power Requirements<\/h3>\n\n\n\n

Separate continuous current from peak current, and define the peak’s magnitude, duration and repetition. Thruster startup, sudden load changes and shifts in propeller loading can push current above the continuous level, so the pack, wiring, connectors and protection devices should be checked against that peak. Assess voltage sag against the vehicle’s minimum allowable bus voltage during peak-load events \u2014 the dynamic behaviour that makes high-power packs sag under transient demand<\/a> worth testing rather than inferring. The path from pack to load adds resistance, so cable cross-section, connector rating and fuse characteristics all count, not just the cell’s discharge rate. A stall draws heavy current only briefly, yet cable and connector heating depends on current squared multiplied by time, which is why the continuous versus peak C-rating<\/a> on a datasheet should be read with both numbers in mind.<\/p>\n\n\n\n

Protection settings deserve the same treatment, because thruster transients are normal events, not faults. If the BMS overcurrent threshold or delay isn’t coordinated with the operating profile, a routine maneuver can trigger a shutdown that looks like a battery failure back on deck. Define the expected peak current, duration and repetition pattern, then check that the BMS protection, motor controller limits and independent circuit protection work together. Soft-start or current limiting can help with startup demand, but suitability depends on the propulsion and control architecture.<\/p>\n\n\n\n

Temperature, BMS, and Fault Protection<\/h3>\n\n\n\n

Low temperatures cut available capacity and raise internal resistance, by an amount that depends on cell design, discharge rate and conditions inside the enclosure. State usable energy at the temperature the vehicle will actually see. There’s a second-order effect too: if the BMS state-of-charge estimate isn’t temperature-compensated, its error grows in cold water and a return-to-surface plan written against nominal capacity stops holding. Characterize the SOC-OCV relationship at low temperature rather than reusing room-temperature data.<\/p>\n\n\n\n

Heat leaves the pack by conduction. In a sealed system it travels from the cells through thermal interface materials, mounting structure, potting and the enclosure wall before reaching the water. How well that path works depends on pack layout, contact between components and enclosure materials, so a pack that behaves in an open-air test may behave differently inside a sealed pressure housing. Thermal validation should reflect the actual installation.<\/p>\n\n\n\n

The BMS monitors operating conditions and manages the protection functions assigned to it. Depending on the architecture, additional overcurrent, short-circuit and insulation-monitoring devices may be required, and its data interface has to suit the vehicle’s control system, whether that is CAN, SMBus or UART. Normal thermal management and fault protection are separate concerns, and protection thresholds should come from the cell and pack specification plus testing, not from one number applied to every design.<\/p>\n\n\n\n

Two conditions sit outside the operating window rather than inside it. Condensation forms whenever the pack is colder than the dew point of the surrounding air, which is routine when a cold pack meets a warm, humid deck; over time it drives insulation loss and terminal corrosion. And an enclosure sealed absolutely, with no relief path, becomes a pressure vessel if a cell vents. Vent paths or defined weak points, humidity monitoring and a temperature-equalization period before recharging belong to the same protection strategy.<\/p>\n\n\n\n

Connectors, Sealing, and Corrosion<\/h3>\n\n\n\n

Connectors must carry the current and voltage, hold mechanically, seal against water and stay serviceable. Seawater intrusion or an exposed conductor can reduce insulation resistance, allow leakage current or create a short path, and how serious that becomes depends on the architecture: an isolated system with insulation monitoring behaves very differently from one tied to the vehicle’s common ground. Subsea practice treats these as qualification items rather than catalogue figures \u2014 standards such as API 17F and ISO 13628-6 for subsea production control systems call for pressure-and-temperature cycling and mate\/demate cycle testing before an interface is accepted.<\/p>\n\n\n\n

Whether the connection ever needs to be broken underwater decides the interface type. Wet-mate connectors are designed to be mated and demated while submerged, which suits servicing or module replacement in the water. Dry-mate connectors are mated in a dry environment and work when the battery stays installed and connected throughout the mission. If underwater disconnection isn’t part of the operating procedure, a wet-mate interface adds cost and integration complexity for no functional gain; the choice should follow the maintenance strategy, connection frequency, depth, current and sealing validation.<\/p>\n\n\n\n

For tether-connected ROVs, strain relief belongs in the same design. Tether tension and handling loads should pass through dedicated mechanical supports rather than the connector contacts or terminations, which otherwise lose contact stability and sealing performance during deployment and recovery.<\/p>\n\n\n\n

Grounding topology should be settled before anodes and materials are chosen. In a floating system, insulation resistance is monitored and an alarm raised before the vehicle keeps descending. In a system tied to common ground, any insulation fault closes a circuit through seawater, so immersed metal corrodes electrolytically and gas forms at the fault. The two topologies shouldn’t be mixed within one vehicle. Anodes deserve particular care: zinc or aluminium anodes protect structural steel, but on a vehicle whose battery negative is tied to common ground, an anode can end up sharing a circuit with that negative, spending itself early and driving stray currents into nearby metalwork.<\/p>\n\n\n\n

Creepage matters at the same stage. Seawater is conductive, so a few millimetres between exposed terminals at operating voltage is a leakage path, and electrochemical migration can grow dendrites across it until a short forms. Size clearance for the highest working voltage, keep connectors dry or fluid-filled, and use humidity or dew-point monitoring inside the enclosure as the earliest warning of ingress.<\/p>\n\n\n\n

Dissimilar metals need the same scrutiny. Electrically connected in seawater, metals at different potentials form a galvanic couple that accelerates corrosion of the less noble one (NPL guide on bimetallic corrosion<\/a>). Insulating sleeves, washers, coatings or a better-matched material pairing can reduce the risk, but the fix has to be evaluated as part of the whole enclosure and grounding design.<\/p>\n\n\n\n

These interface decisions share a theme: pack, cables, connectors and loads form one electrical path, and a fault anywhere along it interrupts power even when the pack itself is healthy.<\/p>\n\n\n\n

Validate the Battery Pack Before Deployment<\/h2>\n\n\n\n

Once the battery requirements have been defined, the next step is to confirm that the proposed pack can meet them under representative operating conditions. For OEM projects, this requires more than reviewing cell datasheets or nominal capacity figures. The pack’s electrical performance, environmental protection and integration details all need to be supported by relevant test data.<\/p>\n\n\n\n

What Battery Data Should OEMs Request?<\/h3>\n\n\n\n

When discussing a battery specification with a manufacturer, OEMs should request documented performance data together with the conditions under which each result was obtained. The essentials are:<\/p>\n\n\n\n