PaperlayerView exact revision on arXiv

Source-linked AI summary

Re-examining rates of lithium-ion battery technology improvement and cost decline

Micah S. Ziegler, Jessika E. Trancik

arXiv:2007.13920v2physics.soc-ph

TL;DR

The paper addresses limited and uncertain evidence for estimating lithium-ion cost and performance improvement, which matters for assessing future adoption and informing policy and technology development. It harmonizes diverse historical data and applies performance-curve models, including a method that incorporates energy density and specific energy. The expanded service definition considerably increases estimated improvement rates, while the authors identify potentially faster cost declines for stationary applications but call for mechanistic modeling to characterize that potential.

  • Problem

    Existing cost-decline models rely on limited data series and measures of technological progress, leaving uncertainty about lithium-ion technologies’ future cost and performance trajectories.

  • Method

    The study systematically collects, harmonizes, and combines data on lithium-ion price, market size, research and development, and performance, then applies performance-curve models with additional performance characteristics.

  • Results

    Incorporating energy density and specific energy considerably increases estimated lithium-ion technological improvement rates, while annual price decreases are estimated at 13.1% for all cells and 13.3% for cylindrical cells.

  • Takeaways & Limitations

    The results suggest previously reported improvement rates may be underestimated and that stationary applications with relaxed volume and mass restrictions might achieve faster cost declines.

  • Takeaways & Limitations

    Source searching and reference reading were primarily conducted in English, with other-language resources translated using online tools.

Abstract

from arXiv · show

Lithium-ion technologies are increasingly employed to electrify transportation and provide stationary energy storage for electrical grids, and as such their development has garnered much attention. However, their deployment is still relatively limited, and their broader adoption will depend on their potential for cost reduction and performance improvement. Understanding this potential can inform critical climate change mitigation strategies, including public policies and technology development efforts. However, many existing models of past cost decline, which often serve as starting points for forecasting models, rely on limited data series and measures of technological progress. Here we systematically collect, harmonize, and combine various data series of price, market size, research and development, and performance of lithium-ion technologies. We then develop representative series for these measures and employ performance curve models to estimate improvement rates. We also develop a method to incorporate additional performance characteristics into these models, including energy density and specific energy performance metrics. When energy density is incorporated into the definition of service provided by a lithium-ion cell, estimated technological improvement rates increase considerably, suggesting that previously reported improvement rates might underestimate the rate of lithium-ion technologies' change. Moreover, our estimates suggest the degree to which lithium-ion technologies' price decline might have been limited by performance requirements other than cost per energy capacity. These rates also suggest that battery technologies developed for stationary applications, where restrictions on volume and mass are relaxed, might achieve faster cost declines, though engineering-based mechanistic cost modeling is required to further characterize this potential.

Broader context

The paper combines historical data and performance-curve models to characterize lithium-ion improvement and its possible drivers. Including energy density and specific energy reveals faster estimated technological change than measures focused only on cost.

  • The study collects and harmonizes data describing lithium-ion technologies, their improvement, and possible drivers of advancement.
  • Performance curve models measure technological change over time and with increasing market size and inventive activity.
  • The proposed method incorporates additional performance dimensions, including energy density and specific energy, into measures of technological change.
  • Incorporating these characteristics suggests that previous measures may have underestimated lithium-ion improvement rates.
  • The methods are intended to help characterize change over time in lithium-ion and other energy- and environmentally relevant technologies.

Introduction

The introduction frames uncertainty about lithium-ion cost trajectories and argues that cost-per-energy-capacity analyses omit important performance improvements. The paper addresses this gap by harmonizing historical data and expanding performance-curve analysis to include energy density and specific energy.

  • Uncertainty about lithium-ion cost and price decline complicates assessment of their future cost competitiveness and broader adoption.
  • Existing analyses commonly relate price per energy capacity to cumulative production, but reported learning rates span 14 to 30%.
  • This range produces widely varying projections for when lithium-ion technologies reach cost or price targets and how much investment is required.
  • Analyses often separate physical improvements such as energy and power packing from cost and price declines, potentially distorting estimated improvement rates.
  • The paper combines historical price, production, and development data, develops representative series, and examines relationships with time, market size, and research and development activity.
  • Adding energy density and specific energy to the service definition considerably increases estimated technological change rates.
  • The results suggest faster cost improvement may be possible for stationary applications with relaxed volume and mass restrictions.

Methods

The study builds a harmonized evidence base from diverse sources, then applies data-processing and regression procedures to model lithium-ion prices and their determinants. It also acknowledges language-related limits in source collection.

  • The researchers collected data from academic, governmental, and business articles, reports, and presentations, tracing sources to original data where possible.
  • They excluded clearly derivative series when underlying data were available, while retaining series that combined reported and otherwise unreported data.
  • The resulting battery database contains 1716 unique lithium-ion and lithium-ion polymer cell records from 1990 through 2019.
  • Energy density and specific energy values were calculated from other reported metrics, with additional calculation details provided in the supplementary information.
  • Price relationships with determinants were modeled using log transformations where appropriate followed by ordinary least squares linear regression.
  • The analysis uses 0.95 prediction intervals alongside trend lines and adjusts currency values for exchange rates and inflation.
  • Limitations: Source searching and reference reading were primarily conducted in English, while other-language resources relied on online translation tools.

Results

Lithium-ion cell prices declined substantially, but estimated improvement and learning rates vary with data choices and service definitions. Including energy density reveals faster technological improvement than energy-capacity measures alone.

  • Prices declined by about 97% since lithium-ion cells’ commercial introduction in 1991, despite exceptions around 1995 and 2008.
  • 4.8–23% spans previously reported annual price-decrease estimates, producing 14–30% learning-rate estimates and projections differing by decades.Projected threshold-crossing dates span nearly 20 years for 75 USD/kWh and nearly 30 years for 20 USD/kWh.
  • 13.1% and 13.3% were the estimated annual price-decrease ratios for all cell types and cylindrical cells, respectively; learning rates were 20.4% and 24.0%.The inventive activity rate estimated from cumulative patent filings was 40.1%.
  • Learning-rate estimates depend on the examined time period and market-size measure, while cumulative cell-count measures produced slightly higher rates than energy-capacity measures.The possible rates encompass nearly all previously reported learning rates, and time-period effects are larger than market-size-unit effects.
  • Energy density rose from approximately 200 Wh/l to over 700 Wh/l, while specific energy rose from approximately 80 Wh/kg to over 250 Wh/kg between 1991 and 2018.Incorporating energy density made trend-line slopes considerably steeper, indicating faster measured technological improvement.

Discussion

Combining 90 harmonized data series yields representative estimates of lithium-ion price and improvement trends. Expanding service beyond energy capacity reveals faster technological change, while data and modeling limitations constrain interpretation and future projections.

  • Data and modeling: 90 harmonized data series describe lithium-ion prices, market size, inventive activity, and performance across cell types.The representative series support comparisons of trends and improvement rates across all cells and cylindrical cells.
  • Price and learning rates: 13.1% and 13.3% annual price decreases were estimated for all cell types and cylindrical cells, respectively.These estimates are similar to the mean and median annual decrease percentages calculated from the collected price series.
  • Drivers of change: A doubling of cumulative patent filings was associated with a 40.1% price reduction, compared with 31% in an earlier estimate.Cumulative market size and inventive activity measures were more strongly correlated with price per service than annual measures.
  • Expanded service definition: Including energy density increased annual price-per-service decreases from 13.1% and 13.3% to 17.1% and 17.4% for all and cylindrical cells.Learning rates likewise increased to 26.6% and 30.9%, indicating that energy-capacity-only measures underestimate technological improvement.
  • Implications and limitations: Stationary-storage requirements may permit faster cost declines because volume and mass restrictions are relaxed, but engineering-based mechanistic modeling is still needed.Historical capacity-fade data are scarce and difficult to compare across applications, limiting retrospective analysis of this potential.

Concluding remarks

The paper provides harmonized data and performance-curve estimates of lithium-ion technological advancement, extending price-per-energy-capacity measures with additional performance characteristics. Including energy density or specific energy substantially increases estimated improvement rates and suggests stationary applications may enable faster cost declines, although mechanistic modeling is still needed.

  • Concluding remarks: The study combines price, market-size, inventive-activity, and technical-performance data to estimate lithium-ion advancement rates.The authors describe systematic data detailing and performance-curve estimates as the basis for their analysis.
  • Concluding remarks: Prices of cylindrical cells and all cell types declined similarly over time, but cylindrical cells showed a considerably higher learning rate.The comparison distinguishes time-based price decline from learning rates associated with market experience.
  • Concluding remarks: Energy density or specific energy included in the service definition considerably increases estimated improvement rates.The paper proposes incorporating these additional cell attributes into the definition of service provided.
  • Concluding remarks: Additional performance characteristics indicate that price-per-energy-capacity measures might underestimate how rapidly lithium-ion technologies improved.The increase in improvement rates also indicates how much price decline may have been limited by performance requirements beyond cost per energy capacity.
  • Concluding remarks: Relaxed volume and mass requirements in stationary storage may allow cost or price for a different service to decline more rapidly.The paper links this possibility to changing research, development, and production priorities as those performance requirements are relaxed.
  • Concluding remarks: Engineering-based mechanistic modeling is required to further evaluate the potential for faster lithium-ion cost declines.The authors also frame their methodology as applicable to broader cost or price-per-service metrics and projections of technological change.

Notes

Performance curves describe relationships between a performance measure and an experience measure. The paper uses this general term for such relationships and related forms.

  • Notes: A performance curve relates a performance measure to an experience measure.
  • Notes: These relationships connect observed technological performance with an experience measure used in the analysis.
  • Notes: The paper uses “performance curves” as the general name for these relationships and others like them.
Loading 2007.13920v2…