AGV Battery Selection Guide: Voltage, Capacity & Charging Requirements
In the past three years, we've seen more AGV projects get battery selection wrong than right. Not because clients lack technical knowledge, but because the spec sheets suppliers provide are basically useless for making real decisions. Two batteries both labeled 48V 100Ah, one costs three times more than the other. What's the difference? That's what this guide is about.
Voltage Selection
Your AGV vendor says the vehicle needs 48V, so you go shopping for 48V batteries. Seems straightforward. It's not.
Basic physics first.
Power equals voltage times current. To output 5000W, a 48V system pulls 104A, a 36V system pulls 139A, a 24V system pulls 208A. Higher current means thicker cables, faster contactor wear, more heat in connectors. Using 24V where you could use 48V is asking for trouble down the road.
But real projects aren't that simple. We did a cold storage project last year, originally spec'd 48V LFP. During testing, internal resistance spiked in cold temps, charging protection kept triggering. Switched to 36V. Less efficient on paper, but reliable. In freezer environments, reliability beats efficiency every time.
My rules for voltage selection:
24V
24V works for sub-500kg loads, under 2kW power demand, tight budgets. Honestly we rarely recommend 24V anymore unless clients specifically request it.
48V
48V covers most warehouse AGVs. Handles 1 to 2.5 ton loads, mature charging infrastructure, widest supplier selection. Default choice if nothing special going on.
72V+
72V and 80V for heavy forklifts, port equipment, outdoor haulers. Samsung SDI built a 94V system for 10-ton electric forklifts that supposedly matches diesel performance (that's from their website, haven't tested it myself, but Samsung's track record in this space is solid).
Capacity: Everyone Knows the Formula, Nobody Knows Where to Get the Inputs
The Formula
Capacity (Ah) = Power (W) × Runtime (hours) ÷ Voltage (V) ÷ Efficiency
Efficiency: 0.95 for lithium, 0.85 for lead acid.
The problem is the "power" number. AGV vendors give you "typical conditions" meaning flat floors, standard loads, no frequent starts and stops. Real warehouses have ramps, overloading, peak hours with continuous operation. Actual power draw can run 50% higher than spec sheet numbers.
What we do: ask the AGV vendor for three numbers. Unloaded cruise power, loaded uphill power, peak power. Then estimate weighted average based on the client's actual routes. If the vendor can only give one vague "average power" number, I multiply by 1.3 before calculating.
For reference, JD Logistics published data on their Asia No.1 smart warehouses. Their Dilang AGVs carry 500kg, run 6 to 9 hours loaded on imported cells, support 50A fast charging that completes in 30 minutes (source: JD Logistics technical documentation). At 48V, 50A charging means battery capacity somewhere between 40Ah and 60Ah. That's reasonable for a 500kg carrier. If your vehicle carries more or runs longer routes, scale capacity up accordingly.
Chemistry: LFP, NMC, LTO Each Have Their Place
Three lithium chemistries dominate the AGV market. Here's my take:
LFP (lithium iron phosphate)
Right choice for 90% of projects. Long cycle life, good safety profile, costs have come down significantly. CATL and BYD make reliable LFP cells with mature supply chains. Unless you have a specific reason to choose something else, default to LFP.
NMC (nickel manganese cobalt)
Higher energy density than LFP, but shorter cycle life and higher thermal runaway risk. Consider NMC only if your AGV battery compartment is too small to fit the LFP volume you need. Personally I don't like NMC in warehouse environments. If thermal runaway happens, you lose the whole warehouse worth of inventory, not just the AGV.
LTO (lithium titanate)
Special case. Energy density is half of LFP (really low), but charging speed is extreme (5C is no problem, full charge in 12 minutes), cycle life is extreme (15000+ cycles), cold weather performance is extreme (works at minus 30°C). Price is also extreme. Two scenarios where LTO makes sense: continuous 24/7 operation with only 10 to 15 minute charging windows, or cold chain environments with serious sub-zero temperatures. If your project fits either scenario, LTO's premium price is justified.
| Type | Energy Density | Cycles | Specs |
|---|---|---|---|
| LFP | 90 to 160 Wh/kg | 2000 to 5000 | Handles 1C charging fine, works down to minus 20°C, moderate cost. |
| NMC | 150 to 220 Wh/kg | 1000 to 2000 | Be careful above 1C charging, average cold performance, higher cost. |
| LTO | 50 to 80 Wh/kg | 15000 to 20000 | Handles 5C charging, works at minus 30°C, significantly higher cost. |
Same LFP Chemistry, Different Capacity: How to Choose
People ask this a lot. Most suppliers don't explain it well because they want you to buy bigger (higher margin).
Is bigger always better? No.
Large capacity problems
First, weight. A 100Ah LFP pack weighs around 45kg, 200Ah weighs 85kg. That extra 40kg either cuts into payload capacity or increases energy consumption per trip.
Second, charging time. At 0.5C, 100Ah takes 2 hours, 200Ah takes 4 hours.
Third, if each shift only uses 30% of battery capacity, 70% of those cells sit at high state of charge every day, accelerating calendar aging.
Small capacity problems
Small capacity problems are obvious: short runtime, risk of vehicles dying mid-shift if charging schedule slips.
My sizing principle
Figure out how much energy the vehicle actually consumes between charging opportunities, then select capacity so you're using 60% to 70% of the pack each cycle. Not deep enough to stress the cells, not shallow enough to waste money on unused capacity.
Example: AGV can charge every 4 hours, average power consumption 800W. Four hours at 800W is 3200Wh. At 48V that's 67Ah. A 100Ah pack means 67% utilization per cycle. That's the sweet spot. A 150Ah pack would only use 45% per cycle, wasting capital on cells that aren't earning their keep.
DOD and Cycle Life: This Knowledge Will Save You Money
DOD means Depth of Discharge. Its impact on battery life is bigger than most people realize.
Battery University (batteryuniversity.com) published test data I'll cite directly:
LFP cells cycled at 100% DOD reach 80% capacity around 600 cycles. Same cells cycled at 40% DOD last over 3000 cycles. Same battery, shallow cycling delivers several times more total energy throughput over its lifetime than deep cycling.
What does this mean practically? Better to buy slightly larger capacity and only use half of it than to buy exactly what you need and drain it every cycle. The second approach looks cheaper upfront but you'll replace batteries more often.
Our standard configuration now: SOC floor at 20% to 25%, ceiling at 80% to 85%, vehicles operate within this window. Combined with opportunity charging, battery life extends two to three times compared to full charge/discharge cycling.
One common misconception
Many people think lithium batteries need periodic full discharge to "calibrate." That's nickel cadmium thinking, completely wrong for lithium. Deep discharge damages electrode structure. Modern BMS uses coulomb counting for SOC estimation, doesn't need empty cycles for calibration.
Charging Strategy: Opportunity Charging or Battery Swap
Both approaches have their place. Neither is inherently "more advanced."
Opportunity charging
Works when AGV utilization stays below 75%. Vehicle charges while waiting for tasks, queuing, loading/unloading. Each session runs 15 to 30 minutes. Adds up to enough runtime across the day. No human intervention needed. With LFP supporting 1C charging, 15 minutes adds roughly 25% capacity, enough for another hour or two of operation.
Battery swap
Works when utilization exceeds 95% and vehicles cannot stop. Vehicle enters swap station, battery changed in 2 to 5 minutes, immediately back to work. Trade-off: each vehicle needs 2 to 3 batteries, dedicated swap equipment, operators to perform swaps.
DHL's case is worth noting. Their global distribution centers use Locus Robotics collaborative robots with opportunity charging. Picking efficiency improved 50% to 70%, workplace injuries dropped 15% (source: DHL website and Logistics Viewpoints coverage). Those efficiency gains aren't just from the robots themselves. Opportunity charging enables true 24-hour unmanned operation. Labor cost savings are substantial.
Cold Environment Pitfalls
If your warehouse has refrigerated or freezer zones, battery selection requires extra care.
Lithium batteries have two problems in cold. First, capacity drops: at minus 10°C, LFP retains maybe 85% capacity; at minus 20°C maybe 70%. Second, charging restrictions: charging lithium below freezing causes lithium plating on the anode. That's permanent damage plus safety hazard. Most BMS refuse charge current when cell temperature drops below 0°C.
Practical solutions: either set up charging zones outside the cold area (vehicles exit to charge, return when done), or use battery systems with preheating (warm cells above freezing before accepting charge current).
If your AGVs work extended periods below minus 25°C, standard LFP probably won't cut it. Consider LTO or specialized cold chain batteries. SVOLT's LMFP cells claim operation at minus 30°C. They've built a decent position in cold chain logistics. Worth investigating for freezer applications.
Running the Numbers
Concrete figures. One 48V AGV, two-shift operation (16 hours daily), five-year horizon.
Lead acid approach
Batteries (2 for rotation):$4,000
Charger:$1,500
Annual electricity (5 yrs):$9,000
Annual maintenance (5 yrs):$3,000
Replacement (Year 3):$4,000
Downtime losses (5 yrs):$12,000
Five-year:total$33,500
LFP approach
Battery (only one):$12,000
Fast charger:$3,500
Annual electricity (5 yrs):$5,750
Maintenance:$0
Replacement:$0
Downtime losses (5 yrs):$2,000
Five-year:total$23,250
Savings: $10,250, roughly 31%
This doesn't even count lead acid disposal costs or lithium residual value after five years.
When does lithium not make sense? Single shift operation (under 8 hours daily), low utilization. Lead acid's lower upfront cost takes 7 to 8 years to overcome. If your project horizon is only 3 to 5 years and you really only run single shift, lead acid might pencil out better.
BMS: More Than Protection Circuits
Many people treat BMS as just protection circuits. Cut off the battery during overcharge, over-discharge, over-current, over-temperature. Job done. Modern BMS does much more.
SOC estimation
Good BMS achieves accuracy within plus or minus 1% to 3% (Infineon's technical docs claim plus or minus 1% for their solutions). Why does accuracy matter? Your dispatch system can assign tasks based on each vehicle's actual remaining charge. High-charge vehicles get long-distance tasks, low-charge vehicles get short runs or head to charging. If SOC estimation error is 10%, this optimization becomes impossible.
SOH monitoring
State of Health tracking. Good BMS tracks capacity degradation curves for each battery, warns months ahead when a pack is failing. Better than waiting for a dead vehicle and dealing with emergency downtime.
Temperature management
Real-time monitoring across multiple points in the pack. Spots hot zones early, reduces power before problems escalate.
Communication interface
CAN bus or RS485 connecting to your warehouse management system. Lets WMS see battery status for every vehicle.
Don't cheap out on BMS. We've seen clients save a few hundred dollars on budget BMS units, then discover SOC estimation errors so large the dispatch system can't use the data at all. Might as well not have the feature.
Certifications: Not Just Paperwork
Two certifications matter most: UN38.3 and IEC 62619.
UN38.3
Covers transport safety. Tests altitude simulation, temperature cycling, vibration, shock, short circuit, overcharge, forced discharge. Required for any lithium battery crossing borders. No UN38.3, logistics companies won't touch it.
IEC 62619
Covers industrial lithium battery use safety. Test items overlap with UN38.3 but focus more on operational scenarios: nail penetration, drop testing, BMS function verification. IEC 62619 forms the technical basis for CE marking in Europe, AS/NZS certification in Australia, PSE approval in Japan.
Important point: certifications apply to specific product models, not to manufacturers. Some manufacturers certify one model then slap the certificate on everything they sell. When purchasing, verify the certificate model number matches exactly what you're buying. Mismatched certificates are worthless.
Supplier Notes
A few we've dealt with:
Among Chinese suppliers, BYD holds major market share in electric forklift lithium batteries, reportedly around 60% in China. Their blade batteries come with 10-year warranty. CATL and Hangcha formed a JV called Pengcheng New Energy specifically for forklift and AGV batteries. If you're already using Hangcha vehicles, that channel might be smoother.
Among Western suppliers, EnerSys NexSys series is solid. Their NexSys AIR wireless charging system supports lead acid, TPPL, and lithium on the same infrastructure. Useful if your fleet is transitioning from lead acid to lithium and you want to protect your charging equipment investment.
Questions to ask when evaluating suppliers:
Which cell manufacturer, what grade (A-grade or B-grade)? Can they provide cycle life test data for packs, not just bare cells? Does warranty cover capacity fade or only complete failure? How do they handle warranty claims (shipping batteries internationally is expensive and slow)?
Final Thoughts
Battery selection isn't a one-time purchase decision. What you choose today determines operating costs and maintenance complexity for the next 5 to 10 years. Don't just look at initial purchase price. Calculate five-year TCO before deciding.
If you have a specific project in the works and need to validate your battery spec, send over the operating parameters, budget range, special requirements (cold storage, explosion-proof, unusual dimensions). We've stepped on some of these landmines before and can help you avoid them.
