Series vs. Parallel Circuits
How real battery packs, grid-scale BESS installations, solar arrays and wind collector systems are actually wired — and why the wiring topology determines everything from failure modes to fault tolerance.
Musk's Rocket Science · Part 1 — Electricity Basics · Chapter 5 · 18 min read
No real energy system is built from a single cell, a single solar module, or a single turbine. A utility-scale Battery Energy Storage System might contain hundreds of thousands of individual cells. A solar farm might have tens of thousands of PV modules. An offshore wind farm might have 80 or more turbines. The engineering problem in every one of these systems is the same: how do you combine many identical building blocks into one coherent, high-power system? The answer is always some combination of two wiring topologies — series and parallel — and the choice between them is never arbitrary.
This chapter treats series and parallel wiring the way a systems engineer actually encounters it: not as an abstract circuit diagram, but as the literal architecture of a battery pack, a BESS rack, a solar string and a wind farm collector network.
5.1 — Two fundamental ways to combine sources
Series and parallel connections are governed by two conservation principles first formalised by Gustav Kirchhoff in 1845. Kirchhoff’s Current Law (KCL): charge cannot accumulate at a junction, so current in must equal current out. Kirchhoff’s Voltage Law (KVL): the sum of voltage rises and drops around any closed loop must equal zero. These two laws are the formal reason series and parallel circuits behave the way they do.
Table 5.1 — Series and parallel, formally
| Series | Parallel | |
|---|---|---|
| Physical arrangement | End-to-end, single current path | Side-by-side, sharing common terminals |
| Current (I) | Same through every element (KCL — only one path) | Splits across branches, sums at the junction |
| Voltage (V) | Adds across elements (KVL) | Same across every branch |
| Resistance | R_total = R₁ + R₂ + … + Rₙ | 1/R_total = 1/R₁ + 1/R₂ + … + 1/Rₙ |
V_total = V₁ + V₂ + … + Vₙ, I_total = I₁ = I₂ = … = Iₙ
In series, voltages add and the current is common to every element.
I_total = I₁ + I₂ + … + Iₙ, V_total = V₁ = V₂ = … = Vₙ
In parallel, currents add and the voltage is common to every branch.
5.2 — Series-connected cells: building pack voltage
Every LiFePO4 e-rickshaw pack is fundamentally a series string. A single cell delivers a nominal 3.2V — far too little to drive a traction motor efficiently. Wiring cells in series along a single current path lets their voltages add, while the current through every cell in the string remains identical at every instant.
Worked example 5.1 — Building a 51.2V pack
Sixteen LiFePO4 cells, each 3.2V nominal, wired in series — a “16S” configuration:
V_pack = 16 × 3.2V = 51.2V
The same current flows through all sixteen cells simultaneously. If the pack delivers 100A to the motor controller, every single cell in the string is also carrying exactly 100A, regardless of its individual state of health.
5.2.1 — The weakest-link problem in series systems
Because current is identical throughout a series string, a weak cell cannot simply do less work — it is forced to carry the same current as every healthy cell around it. Recall from Chapter 3 that a weaker cell typically has higher internal resistance, so V = I·r means that cell sags in voltage more than its neighbours under the exact same current.
Pushed far enough, that one weak cell reaches its under-voltage protection limit long before the rest of the string is anywhere near depleted. And because the BMS must protect the whole string, the entire pack’s usable capacity becomes limited by its single weakest cell.
Technical framing
This is precisely why active and passive cell balancing exist. Without periodically equalising individual cell voltages within a series string, manufacturing tolerance and uneven ageing inevitably let one cell become the limiting factor for the entire pack’s usable energy, well before the other fifteen cells are exhausted — which is what our active balancing and capacity test procedure is designed to measure.
5.3 — Parallel-connected strings: building capacity in a BESS rack
A single 16S string at 51.2V and 100Ah stores about 5.12kWh — nowhere near enough for grid-scale storage. To scale up energy while holding a fixed, manageable bus voltage, BESS designers connect multiple identical strings in parallel, all sharing a common positive and negative bus.
Worked example 5.2 — Sizing a BESS rack
Sixteen identical 51.2V / 100Ah strings, connected in parallel to a common 51.2V DC bus:
Bus voltage stays fixed: V_bus = 51.2V (same as a single string)
Total capacity: 16 × 100Ah = 1,600Ah at 51.2V
Total energy: 16 × 5.12kWh = 81.92 kWh
Enough to power a typical Indian household for several days, from a single BESS rack.
Important
Parallel strings must be closely matched in voltage and internal resistance before being connected together. If one string sits at a meaningfully higher voltage than its neighbours at the moment of connection, a large, uncontrolled circulating current can flow between strings even with no external load at all — since parallel branches must by definition settle to the same voltage (KVL). This is why BESS integrators pre-charge and voltage-match strings before closing parallel contactors.
5.4 — Series-parallel notation and the real architecture of a BESS
Industry shorthand describes a pack’s topology as mSnP — m cells in series, n of those strings in parallel. A single 16-cell string is 16S1P; the BESS rack above is a 16S16P configuration: sixteen cells in series, sixteen such strings in parallel.
Table 5.2 — BESS scale hierarchy (illustrative, representative of commercial systems)
| Level | Typical configuration | Voltage | Energy |
|---|---|---|---|
| Cell | Single LiFePO4 prismatic cell | 3.2V | ~0.32 kWh |
| Module | 16S1P | 51.2V | ~5.1 kWh |
| Rack | 16S16P | 51.2V bus | ~82 kWh |
| Container | Multiple racks on a common high-voltage DC bus | ~1,000–1,500V DC | ~2–4 MWh (Megapack-class, illustrative) |
Why this matters later
Notice that the container-level bus voltage of roughly 1,000–1,500V DC is far higher than a single rack’s 51.2V — achieved by connecting racks in series, not parallel, at the container level, exactly as individual cells were series-connected to build the original 51.2V module.
The same series-for-voltage, parallel-for-capacity logic repeats at every scale, from a single cell up to a grid-connected container.
5.5 — Solar PV: strings and arrays
Solar power systems face an almost identical engineering problem. A single photovoltaic module might produce an open-circuit voltage (V_oc) around 40–50V — far below the several-hundred to thousand-volt input window a modern string inverter’s Maximum Power Point Tracker (MPPT) needs for efficient operation. The solution, again, is series connection.
Worked example 5.3 — Sizing a solar string
A commercial PV module rated V_oc ≈ 49.5V. A string inverter’s MPPT operating window is roughly 200–850V DC.
V_string = 20 × 49.5V = 990V
This sits comfortably within the inverter’s DC input window — but close to many inverters’ 1,000V or 1,500V absolute maximum rating. Designers must also account for the fact that V_oc rises in cold weather, a real temperature-coefficient effect, and size the string with margin below the absolute maximum even on the coldest expected morning.
Just as a BESS scales capacity by paralleling strings at a fixed bus voltage, a solar array scales total power by connecting many identical strings in parallel at a combiner box, all feeding the same inverter input.
5.5.1 — Partial shading: solar's version of the weak-cell problem
Because a solar string is a series circuit, it inherits exactly the same weakest-link vulnerability as a battery pack. If a single module in a 20-module string is partially shaded — by a cloud edge, a nearby structure, or even bird droppings — that module’s current-producing capability drops sharply. And because all twenty modules must carry the same current (KCL again), the shaded module throttles the current for the entire string, not just its own output.
In plain English
This is functionally identical to the weak-cell problem in a battery string from section 5.2.1: one underperforming series element drags down the output of every healthy element around it.
It is why utility-scale solar sites invest heavily in vegetation management, module cleaning and site layout specifically to minimise inter-row shading, and why module-level power electronics — power optimisers and microinverters — exist as an alternative architecture that breaks the series-current constraint entirely.
5.6 — Wind farms: parallel collector networks
A modern offshore wind turbine — typically rated 6–15MW — generates its own AC power internally and steps it up through a dedicated pad-mounted or nacelle-based transformer, usually to 33kV. Unlike battery cells, individual turbines are not wired in series with each other. Instead, each turbine feeds into a shared collector network: a set of subsea cables — or underground, for onshore farms — connecting many turbines in parallel along a radial feeder, which then runs to a substation for final step-up to transmission voltage.
Table 5.3 — Wind farm collector network structure (illustrative, representative topology)
| Level | Function | Typical voltage |
|---|---|---|
| Turbine output | Individual generator and power converter | ~690V AC internally, stepped up locally |
| Inter-array (collector) cabling | Connects multiple turbines in parallel along a feeder string | 33 kV AC |
| Offshore or onshore substation | Combines multiple collector feeders, steps up further | 132–220 kV AC |
| Export cable / grid connection | Delivers combined farm output to the transmission grid | 132–400 kV AC, or HVDC for very long offshore links |
Technical framing
Because collector feeders are effectively parallel branches converging at a substation, the entire network benefits from the same fault-tolerance property parallel systems generally offer. Losing one turbine, or even one entire feeder string, reduces total farm output roughly in proportion to the lost capacity, rather than taking down the whole array — a direct contrast to a series battery string, where a single failed cell can disable the entire string.
5.7 — Green hydrogen: electrolyser stacks follow the same logic
A single PEM electrolyser cell — the non-Ohmic device from Chapter 3 — operates at only around 1.8–2.2V under load, far too little to be practical at industrial scale on its own. Electrolyser manufacturers solve this with exactly the same architecture as a battery pack: individual cells are stacked electrically in series inside a single pressure vessel (a “stack”), so their voltages add while the same current — and therefore the same rate of hydrogen production per cell — passes through every cell simultaneously.
Worked example 5.4 — Sizing an electrolyser stack and plant
A PEM electrolyser stack uses 130 cells in series, each operating at approximately 2.0V under rated load:
V_stack = 130 × 2.0V = 260V
To reach a green-hydrogen plant’s target production capacity, multiple identical stacks are connected electrically in parallel to a shared rectifier bus — exactly like paralleling BESS strings onto a common DC bus — so that total plant current, and therefore total hydrogen production rate, scales with the number of stacks while each stack still operates at its optimal 260V.
Important
Just as with a series battery string, one underperforming cell inside an electrolyser stack — through membrane degradation or catalyst fouling — is forced to carry the same current as every healthy cell around it, and limits the achievable current density, and therefore hydrogen output, for the entire stack. The weakest-link problem from section 5.2.1 reappears here in a completely different industry, governed by exactly the same physics.
5.8 — Grid transmission and data centres: redundancy through parallel paths
The same fault-tolerance logic scales up to the transmission grid itself. Critical substations are typically connected by multiple parallel transmission lines rather than a single line, specifically so that the loss of any one line — the so-called N-1 contingency grid planning standard — does not interrupt supply, since the remaining parallel lines can carry the full load, within thermal limits, while the failed line is repaired.
Hyperscale data centres apply an almost identical philosophy at the rack level. Critical IT racks are typically fed by two independent parallel power paths, commonly labelled the “A feed” and “B feed”, each traceable back through separate UPS units, separate backup generation, and ideally separate utility connections. Every critical server has dual power supplies, one on each feed, so that the loss of an entire feed causes zero interruption — the surviving feed carrying the full load exactly as a surviving transmission line does after an N-1 event.
In plain English
This is the same principle repeating at every scale: parallel paths mean no single point of failure. It is a direct engineering consequence of Kirchhoff’s Current Law — current simply redistributes across whatever parallel paths remain available, up to each path’s physical current-carrying limit — whether that path is a wind farm collector feeder, a transmission line, or a data centre’s B feed.
Technical framing
Parallel redundancy also appears one level deeper, inside the power electronics itself. A single SiC or GaN power semiconductor die has a maximum current rating, so high-current inverters and PCS units — in a BESS, an EV traction inverter, or a data centre UPS — often parallel multiple discrete devices or dies within one module to share current, using the same current-summing logic as any other parallel branch. This works only if the devices are matched closely enough in on-resistance to stop one device hogging a disproportionate share of the total current.
5.9 — Quick reference
Series and parallel cheat sheet
| Series | Parallel | |
|---|---|---|
| What adds | Voltage | Current and capacity |
| What stays fixed | Current | Voltage |
| Failure mode | One weak link limits the whole string | One branch loss reduces capacity proportionally |
| Battery example | 16 cells → 51.2V module | 16 strings → 82kWh rack |
| Other examples | 130-cell electrolyser stack → 260V; 20 PV modules → ~1,000V string | Many wind turbines → shared collector feeder; parallel electrolyser stacks; data centre A/B feeds |
Chapter summary
- ✓Series connections add voltage while forcing identical current through every element. This is how a 51.2V battery module is built from sixteen 3.2V cells, how a solar string reaches an inverter’s MPPT window from many 40–50V modules, and how a 260V electrolyser stack is built from 130 individual 2.0V cells.
- ✓Parallel connections add current and capacity while holding voltage fixed across every branch. This is how a BESS rack scales from 5.1kWh to 82kWh by paralleling identical strings, how a wind farm’s collector network combines many turbines onto one feeder, and how multiple electrolyser stacks scale a plant’s hydrogen output.
- ✓Series systems inherit a weakest-link vulnerability: one underperforming element — a weak battery cell, a shaded solar module, a degraded electrolyser cell — throttles the entire string, because every element is forced to carry identical current.
- ✓Parallel systems offer inherent fault tolerance: losing one branch — a failed BESS string, a downed turbine, a tripped transmission line, a data centre’s A or B feed — reduces capacity proportionally rather than disabling the whole system, provided branches are properly matched and protected against circulating currents.
- ✓That same fault-tolerance principle repeats one level deeper inside power electronics themselves, where discrete SiC and GaN devices are paralleled within inverters and PCS modules to share current.
- ✓Real systems at every scale — from a single battery module to a gigawatt-hour grid installation — are built by repeatedly applying series (for voltage) and parallel (for capacity) connection in combination, described industrially with mSnP notation.
Test your understanding
- 1A BESS integrator wants a 400V DC bus using 3.2V LiFePO4 cells. Roughly how many cells must be connected in series? (Use nominal voltage.)
- 2If that same 400V string has a capacity of 280Ah, and the integrator connects 8 identical strings in parallel, what is the total rack capacity in Ah, and the total energy in kWh?
- 3Explain, using Kirchhoff’s Current Law, why one severely underperforming cell in a series string can limit the usable capacity of the entire string.
- 4Why do parallel BESS strings need to be voltage-matched before their contactors are closed?
- 5A wind farm’s collector feeder loses one turbine to a fault. Contrast what happens to farm output with what would happen if a single cell failed open-circuit in a series battery string.
- 6An electrolyser stack uses 130 cells in series at about 2.0V each. If a plant needs to double its hydrogen production rate while keeping each stack at its optimal operating voltage, should it add more cells per stack, or more stacks in parallel? Explain why.
- 7Challenge: a solar string of 24 modules (V_oc = 48V each at standard test conditions) is installed where cold mornings can raise each module’s V_oc by up to 8%. What is the string’s worst-case open-circuit voltage, and why does this matter for inverter selection?
Frequently asked questions
What is the difference between wiring cells in series and in parallel?+
In series, cells are connected end-to-end along a single current path: their voltages add while the same current flows through every cell. In parallel, cells or strings share common terminals: the voltage stays the same across every branch while the currents add. Series builds voltage, parallel builds capacity. Both follow directly from Kirchhoff’s Voltage and Current Laws.
What does 16S16P mean in a battery pack?+
It describes the topology: 16 cells wired in series to form a string, and 16 of those strings wired in parallel. With 3.2V LiFePO4 cells, the 16 in series give a 51.2V bus, and the 16 parallel strings at 100Ah each give 1,600Ah — about 82kWh in total, all at a fixed 51.2V.
Why does one weak cell limit an entire battery pack?+
Because current is identical throughout a series string, a weak cell cannot do less work — it carries the same current as every healthy cell. Having higher internal resistance, it sags further in voltage under that current and hits its under-voltage protection limit first. The BMS must protect the whole string, so the pack’s usable capacity is set by its single weakest cell. This is what cell balancing exists to prevent.
Why must parallel BESS strings be voltage-matched before connection?+
Parallel branches must settle to the same voltage by Kirchhoff’s Voltage Law. If one string sits at a meaningfully higher voltage than its neighbours when the contactor closes, a large uncontrolled circulating current flows between strings — even with no external load connected. BESS integrators pre-charge and voltage-match strings before closing parallel contactors for exactly this reason.
Why does shading one solar panel reduce the whole string’s output?+
A solar string is a series circuit, so every module must carry the same current. A shaded module’s current-producing capability drops, throttling the current for the entire string rather than just its own output — the same weakest-link problem as a weak battery cell. Bypass diodes limit the damage by routing current around the worst-affected cells, and module-level power electronics such as optimisers and microinverters avoid the constraint entirely.
Why are wind turbines wired in parallel rather than in series?+
So that a single failure does not take down the array. Each turbine generates AC internally, steps up locally to about 33kV, and feeds a shared collector network as a parallel branch. Losing one turbine, or even a whole feeder string, reduces farm output roughly in proportion to the capacity lost — in direct contrast to a series battery string, where a single failed cell can disable the entire string.
Reviewed by
Musk's Rocket Science is an original educational series explaining the physics and chemistry behind everyday electricity and battery technology. Figures and worked examples use standard physical constants and representative real-world values for illustration.