Physics-based battery simulation, from particles to cells.
We connect particle-scale electrochemistry to full-cell multiphysics, and use it to design fast-charge protocols, predict ageing, and build the simulation and data workflows your engineers work in every day.
One parameter set, three scales一套参数，三个尺度
PARTICLE → ELECTRODE → CELL
Particle level
Solid-state diffusion, phase transitions, particle cracking and interfacial kinetics — used to screen active materials on mechanism, not correlation.
颗粒尺度
固相扩散、相变、颗粒开裂与界面反应动力学。用于从机理判断活性材料与配方方向。
Electrode level
Porosity, tortuosity, coating thickness and calendering — how process choices set rate capability and the lithium-plating window.
电极与极片尺度
孔隙率、曲折度、涂布厚度与压实密度，如何决定倍率性能与析锂窗口。
Cell level
Coupled electro-thermal-mechanical models: tab and cooling design, temperature rise, and performance over real duty cycles.
电芯级多物理场
电-热-力耦合模型：极耳与散热设计、温升，以及真实工况下的性能表现。
The three scales share one parameter set: values calibrated at the particle scale carry into the electrode and cell models, so a change in formulation or process stays traceable through to cell-level performance.
三个尺度共用一套参数体系：颗粒尺度标定的参数直接进入电极与电芯模型，配方或工艺的改动可以一路追溯到电芯性能。
Services服务
ENGAGEMENTS — PROJECT, RETAINER, OR EMBEDDED
Particle-level modelling颗粒尺度建模
S-01 · CASE 05Diffusion and stress models, phase-change pathways, SEI growth kinetics for mechanism-level material assessment.
扩散与应力模型、相变路径、SEI 生长动力学，用于材料的机理级评估。
Electrode & coating simulation极片与涂层仿真
S-02 · CASE 04Microstructure-resolved and homogenised models linking loading, porosity and calendering to rate and plating risk.
微观结构与均质化模型，将面容量、孔隙率、压实与倍率、析锂风险联系起来。
Cell-level multiphysics电芯级多物理场
S-03 · CASE 06Electro-thermal-mechanical models of pouch, cylindrical and prismatic cells under real duty cycles.
软包、圆柱与方形电芯在真实工况下的电-热-力耦合模型。
Fast-charge protocol design快充协议开发
S-04 · CASE 01Charge current profiles optimised against plating and temperature limits, then validated on your cells.
在析锂与温度边界约束下优化充电电流曲线，并在客户电芯上验证。
Ageing & lifetime prediction老化与寿命预测
S-05 · CASE 02Degradation-mode models with knee detection, calibrated on your cycling data and extrapolated to field duty cycles.
基于衰减机理的模型与拐点识别，用客户循环数据标定，并外推到实际工况。
MCP servers for battery data电池数据 MCP 服务
S-06 · CASE 03AI access to your own test data, so engineers ask questions in plain language. See section 03.
让 AI 接入你们自己的测试数据，工程师用自然语言提问。详见第 03 节。
Data platforms & automation数据平台与自动化
S-07Cycler-file ingestion, feature extraction and reporting pipelines that replace manual spreadsheet work.
充放电机数据接入、特征提取与自动报告流水线，替代手工表格分析。
Consulting & training咨询与培训
S-08Model reviews, DoE planning, and hands-on courses for in-house simulation teams.
模型评审、试验设计规划，以及面向客户内部仿真团队的实操培训。
Ask your cells a question向你的电芯提问
MCP SERVER — BUILT ON YOUR OWN MEASURED DATA
It lets a battery engineer ask questions about cells in plain English and get answers computed from the actual measured data.
让电池工程师用自然语言提问，答案全部由真实测量数据计算得出。
The information needed to judge a cell is scattered — raw cycler files in S3, a measurement database, a tracker spreadsheet, and the Quality team's workbook. Answering “is this cell aging badly, and why?” used to mean opening four systems and doing the analysis by hand.
判断一颗电芯所需的信息是分散的：S3 里的原始循环文件、测量数据库、跟踪表格，以及质量团队的工作簿。要回答“这颗电芯老化是否异常、原因是什么”，过去要打开四个系统并手工分析。
It reads the raw test files and interprets them: segments them into meaningful events (formation, cycling, rate tests, pulses), then describes what happened in engineering language — where capacity was stable, where it started declining, whether there's a knee, whether the impedance is growing faster than the capacity is fading. It cross-links that to how the cell was built and what the Quality inspections found, and it can compare a cell against similar ones from history.
它读取原始测试文件并加以解释：先切分为有意义的事件（化成、循环、倍率测试、脉冲），再用工程语言描述发生了什么——容量在哪一段稳定、从哪里开始衰减、是否出现拐点、阻抗增长是否快于容量衰减。它把这些与电芯的制造信息、质量检验结果交叉关联，并可与历史上相似的电芯做对比。
A short written brief in chat — decision-ready sentences, not data dumps — and an interactive dashboard for looking at the curves.
一是对话里的简短结论——可直接用于决策的句子，而不是数据堆叠；二是用于查看曲线的交互式看板。
R&D engineers asking about cell performance, and the Quality team watching process control (Cpk, control charts, defects, lot traceability).
研发工程师关注电芯性能；质量团队关注过程控制（Cpk、控制图、缺陷、批次追溯）。
The numbers are always computed by code, and the model only narrates them. That keeps the answers auditable — you can trace any figure back to measured data.
所有数字都由代码计算，模型只负责叙述。因此每一个数字都可以追溯回原始测量数据。
Read the deployment case: four systems, 31 users, 3.2 h → 6 min →查看落地案例：四个系统、31 位用户、3.2 小时 → 6 分钟 →
How we validate验证方法
EVERY MODEL SHIPS WITH ITS BENCHMARK
Parameterisation
Parameters come from your own measurements, or from a documented literature source with its uncertainty stated.
参数化
参数来自客户自有测量，或来自有出处的文献值，并明确标注不确定度。
Held-out validation
Calibration and validation datasets are kept separate. We report error against data the model never saw.
留出验证
标定与验证数据集严格分开，误差指标基于模型未见过的数据。
Stated limits of validity
Each model comes with the C-rate, temperature and SoC window in which it was validated — and where it should not be trusted.
明确适用边界
每个模型都给出经过验证的倍率、温度与 SoC 区间，以及不应外推的范围。
Handover you can run
Deliverables include the model, the scripts, and a report your team can reproduce line by line.
可复现的交付
交付内容包括模型、脚本与报告，客户团队可以逐步复现结果。
Selected work案例研究
CASE STUDIES — CLIENT NAMES WITHHELD案例研究 — 客户名称已隐去
10→80 % in 22 minutes, with no added plating risk10→80 % 充电 22 分钟，析锂风险不增加
NMC811 21700 cell. A plating-aware model and reference-electrode cells gave a four-step profile that cut charge time by a third, with the same cycle life.
NMC811 21700 电芯。基于析锂判据的模型加三电极电芯验证，得到四段式充电曲线：充电时间缩短三分之一，循环寿命不变。
Knee onset predicted from the first 300 cycles用前 300 次循环预测衰减拐点
96 LFP 280 Ah cells. A degradation-mode model predicted where fade turns non-linear, tested on 32 cells it never saw.
96 颗 LFP 280 Ah 电芯。基于衰减机理的模型预测容量衰减何时转为非线性，并在 32 颗从未见过的电芯上检验。
Four data systems, one question box四个数据系统，一个提问框
A cell R&D centre's cycler files, measurement DB, build tracker and QA workbook, joined behind one MCP server used by 31 engineers.
某电芯研发中心的循环数据、测量数据库、制造跟踪表与质量工作簿，统一接入一个 MCP 服务，31 位工程师日常使用。
A thicker cathode that keeps its 2C rate加厚正极，2C 倍率不打折
NMC622 pouch cell. CT-measured tortuosity and a two-layer porosity design took loading from 3.4 to 4.2 mAh/cm² without losing rate.
NMC622 软包电芯。基于 CT 实测曲折度与双层孔隙梯度设计，面容量从 3.4 提升到 4.2 mAh/cm²，倍率性能基本保持。
Two cathode candidates ruled out before a single cell was built在造电芯之前，先排除两个正极候选
A chemo-mechanical particle model ranked four single-crystal NMC candidates by cracking risk. SEM after 300 cycles confirmed the order.
颗粒级化学-力学模型按开裂风险对四种单晶 NMC 候选材料排序，300 次循环后的 SEM 截面结果与排序一致。
In-cell temperature spread cut from 11.4 K to 4.2 K电芯内部温差从 11.4 K 降到 4.2 K
75 Ah pouch cell for commercial vehicles. A 3D electro-thermal model traced the hot spot to current crowding at same-side tabs.
商用车 75 Ah 软包电芯。三维电-热耦合模型定位到热点来自同侧极耳处的电流集中。
Send us a cell and a question.
Tell us the cell chemistry, the format, and what you need to decide. We reply with the modelling approach we would use, the data we would need from you, and a scope.
带上一个电芯和一个问题联系我们。
告诉我们体系、形状与需要做的决策。我们会回复建议的建模路径、需要你们提供的数据，以及项目范围。