Battery storage: derating, drifting state estimates, wide margins

Batteries and power electronics derate early to stay safe

What happens. Batteries and inverters reduce power when they get warm. The limit is set against a temperature that is measured at one place or assumed for the worst case, so the system backs off long before the cells or the semiconductors are in danger.

What it costs. Power and capacity that the customer paid for and never gets in hot weather or under heavy load, and complaints about a system that cannot deliver its data-sheet figures.

Why the usual fix fails. Putting in a larger cooling system adds cost and own consumption. Moving the threshold up without knowing the inner temperature risks damage and warranty claims. More temperature sensors still show surface points.

How a fixed-price remote diagnosis checks it. From operating logs (currents, voltages, housing and coolant temperatures, derating events) I estimate how the inner temperatures develop and how much margin is left when the system derates. The fixed-price remote diagnosis tells you whether computed inner temperatures can release the real margin, and what that needs.

Background: Computing unmeasured states for control

State of charge is estimated poorly at low currents

What happens. The charge level of a cell is computed from voltage and current. At low currents and in flat parts of the voltage curve, voltage says little and current measurement errors add up. The displayed level jumps or drifts and is corrected at the next full charge.

What it costs. Usable capacity left unused as a reserve, users who do not trust the display, and operating strategies built on a wrong number.

Why the usual fix fails. A better current sensor costs money and still drifts. Frequent full charging to resynchronise is not always possible and wears the cells. A fixed table cannot follow ageing and temperature.

How a fixed-price remote diagnosis checks it. From long logs of voltage, current and temperature, with reference points where you have them, I look at how far the estimate drifts and under which conditions. The fixed-price remote diagnosis tells you whether a better state estimation closes the gap without new hardware.

Background: Computing unmeasured states for control

State of health is known only from tests

What happens. How much a battery has aged is found in capacity tests, which take the system out of service. In operation the figure is a rough guess from counted cycles and time. The warranty promise, though, is given in numbers of capacity.

What it costs. Warranty risk without a reliable picture, tests that interrupt operation, and unfair decisions on claims.

Why the usual fix fails. Cycle counters ignore how hard the battery was used. Scheduled tests cost income. A pessimistic ageing curve over-provisions the system.

How a fixed-price remote diagnosis checks it. From operating data of systems of different ages I look at which signals carry the ageing and how well health can be computed during normal operation. The fixed-price remote diagnosis tells you whether a health estimate in operation is accurate enough to support warranty and service decisions.

Background: Computed measurements

Cell strings drift apart and capacity is lost

What happens. In a large storage system many cells and strings sit in parallel and in series. They differ in capacity, resistance and temperature, and they drift apart over time. The weakest string decides when the system must stop charging or discharging.

What it costs. Capacity that is paid for and cannot be used, uneven ageing and early replacement of single modules.

Why the usual fix fails. Passive balancing in the module burns the extra energy as heat and works slowly. Replacing weak modules is expensive. Measuring each cell individually is costly and still does not predict the drift.

How a fixed-price remote diagnosis checks it. From string voltages, currents and temperatures I look at how the imbalance develops and which strings pull the system down. The fixed-price remote diagnosis tells you whether the imbalance per string can be estimated and the operation adjusted to it.

Cooling runs at full power and eats the system's own energy

What happens. The cooling of a storage system runs at a fixed high level whenever the system works hard, or switches on at a threshold. It does not know what the load will do in the next hour.

What it costs. The system’s own consumption, which reduces the efficiency that the customer sees, and wear on pumps and fans.

Why the usual fix fails. Adjusting the threshold for the hottest day wastes energy on all others. A larger heat exchanger moves the cost to the purchase. Reacting after the temperature has risen is always late.

How a fixed-price remote diagnosis checks it. From operating logs (load profile, temperatures, coolant flow, cooling power) I look at how much cooling energy is spent beyond need. The fixed-price remote diagnosis tells you whether predictive thermal control, using the expected load, can save it.

Background: Economic predictive control

Second-life batteries have unknown cell states

What happens. Used vehicle batteries get a second life in stationary storage. The cells come from different packs with different histories. Nobody knows their real capacity and resistance, and the records are incomplete.

What it costs. Risk of early failure, a reserve that cuts into the business case, and the cost of measuring each module in a test rig.

Why the usual fix fails. Treating all modules as the same average leaves the weak ones to decide. Testing every module separately is slow. Assuming the manufacturer’s data holds after years of use is optimistic.

How a fixed-price remote diagnosis checks it. From the data of a first operation period, such as voltage response to load steps and temperature, I look at whether the state of each module can be estimated during operation. The fixed-price remote diagnosis tells you whether estimation of mixed cells can replace part of the testing.

Vehicle battery cooling follows thresholds, not the route

What happens. Vehicle battery cooling switches on at a temperature threshold. The system does not know that a long climb or a fast-charging stop is coming, so it cools late and then hard.

What it costs. Range lost to the energy spent on cooling, battery temperatures that peak and age the cells, and a driver who sees the car slow down.

Why the usual fix fails. A lower threshold cools earlier and uses more energy all the time. A bigger cooling circuit adds cost and weight. Rules for single cases multiply.

How a fixed-price remote diagnosis checks it. From drive and charging logs with route data I look at how often the current rule is late or early. The fixed-price remote diagnosis tells you whether predictive thermal control with route information would keep temperatures down at lower cooling energy.

Background: Industrial process predictive control

Charge and health estimates in vehicles drift

What happens. The charge level and the health of the battery shown in a vehicle drift over time. After years the remaining range is wrong, and the customer sees a figure that does not fit the experience.

What it costs. Range anxiety, service visits for a pack that is fine, and warranty discussions based on wrong numbers.

Why the usual fix fails. Recalibrating at a workshop is rare. A conservative display hides the problem and gives away usable range. Table-based estimates do not follow ageing and temperature.

How a fixed-price remote diagnosis checks it. From fleet data (voltage, current, temperature, charging events, odometer) I look at how the estimates drift with age and conditions. The fixed-price remote diagnosis tells you whether a better estimate can keep the display correct over the vehicle’s life.

Background: Computing unmeasured states for control

Fast charging is limited conservatively

What happens. The charge current is cut back by tables for temperature and charge level. They protect the cell against lithium plating and heat, and are built for the worst cell in the worst condition.

What it costs. Longer charging time than the battery could take, and a competitive disadvantage on charging time.

Why the usual fix fails. Loosening the table risks cell ageing. A cooling system that cools harder uses energy. Test series for new tables take months.

How a fixed-price remote diagnosis checks it. From charging logs and cell test data I look at how far the charge current could go before the cells show stress, with the state computed in real time. The fixed-price remote diagnosis tells you whether charging at the real limit is realistic.

Background: Economic predictive control

Industrial storage needs control for battery, energy and inverter together

What happens. An industrial battery storage system has to serve the plant: cut load peaks, take in solar power, keep supply during outages. The battery management, the energy management and the inverter control come from different teams and run on different logic.

What it costs. A storage that is not used for what it was bought for, hand-tuned control for each project, and a long commissioning.

Why the usual fix fails. Fixed rules (charge at midday, discharge in the evening) ignore the plant’s real load and the tariff. A control concept per customer is not a product. Bringing in specialists for each project is slow.

How a fixed-price remote diagnosis checks it. From the plant’s load profiles, tariffs and the storage data I look at how much a coordinated, model-based energy management would earn compared with the fixed rules. The fixed-price remote diagnosis tells you whether a common control structure for battery, energy and inverter can serve as a product.

Background: Economic predictive control