''This page is about the intersection of industrial systems and machine learning in the context of climate change mitigation. For an overview of industry as a whole, please see the [https://en.wikipedia.org/wiki/Industry Wikipedia page] on this topic.''[[File:Industry.png|thumb|500x500px|Selected opportunities to use machine learning to reduce greenhouse gas emissions in industry.|alt=]]Industrial production, logistics, and building materialsis area leadingmajor causescause of difficult-to-eliminate GHG emissions<ref>{{Cite journal|last=Davis|first=Steven J.|last2=Lewis|first2=Nathan S.|last3=Shaner|first3=Matthew|last4=Aggarwal|first4=Sonia|last5=Arent|first5=Doug|last6=Azevedo|first6=Inês L.|last7=Benson|first7=Sally M.|last8=Bradley|first8=Thomas|last9=Brouwer|first9=Jack|last10=Chiang|first10=Yet-Ming|last11=Clack|first11=Christopher T. M.|date=2018-06-29|title=Net-zero emissions energy systems|url=https://www.sciencemag.org/lookup/doi/10.1126/science.aas9793|journal=Science|language=en|volume=360|issue=6396|pages=eaas9793|doi=10.1126/science.aas9793|issn=0036-8075}}</ref>, representing over 30% of global GHG emissions in 2010.<ref Fortunatelyname=":1">Fischedick forM., MLJ. researchersRoy, A. Abdel-Aziz, A. Acquaye, J.M. Allwood, J.-P. Ceron, Y. Geng, H. Kheshgi, A. Lanza, D. Perczyk, L. Price, E. Santalla, C. Sheinbaum, and K. Tanaka, 2014: Industry. In: Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Edenhofer, O., R. Pichs-Madruga, Y. Sokona, E. Farahani, S. Kadner, K. Seyboth, A. Adler, I. Baum, S. Brunner, P. Eickemeier, B. Kriemann, J. Savolainen, S. Schlömer, C. von Stechow, T. Zwickel and J.C. Minx (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Available from https://www.ipcc.ch/site/assets/uploads/2018/02/ipcc_wg3_ar5_chapter10.pdf.</ref> The IPCC estimates that energy density can be reduced by up to 25% simply through energy efficiency measures such as replacing and upgrading older equipment, but getting to carbon neutral will require switching carbon-intensive feedstocks, new materials science, improving product life cycles, streamlining supply chains, and even reducing consumer demand.<ref name=":1" /> The global industrial sector – dominated by large firms – spends billions of dollars annually gathering data on factoriestheir own factory operations and and supply chains<ref>Mike Gualtieri, Noel Yuhanna, Holger Kisker, Rowan Curran, Brandon Purcell, Sophia Christakis, Shreyas Warrier, and Matthew Izzi. The Forrester Wave: Big data streaming analytics, Q1 2016. Forrester.com, January 2016.</ref>, –which aidedcan potentially be fed into ML algorithms to help improve production efficiency and decrease carbon-intensive energy use. Such data collection has been facilitated by rapid improvements in thesensors, costautomation technologies, and accessibilityimage ofrecognition, cloud-basedassisted storageby the growing accessibility and computing,cost asof wellcomputing asinfrastructure. sensorsThis andnotion data-gatheringof mechanismsinterconnected suchfactory asequipment, QRreal-time codesdata collection, and imageautonomous recognitionfeedback (vaguelyloops has been referred to variously as [https://en.wikipedia.org/wiki/Fourth_Industrial_Revolution "Industry 4.0" and(or "Industrie 4.0" to recognize its German roots)], [https://www2.deloitte.com/us/en/insights/focus/industry-4-0/smart-factory-connected-manufacturing.html "Smart/Connected Factories"], [https://en.wikipedia.org/wiki/Industrial_internet_of_things "Industrial Internet of Things (IIoT)"], and [https://www.ptc.com/en/blogs/corporate/what-is-a-digital-thread "digital thread"] (connectivity across the supply chain/product life cycle).
By intelligently analyzing these emerging factory and supply chain data, ML can potentially reducelower industrial emissions by helping toindustry streamlinein supplythe chains,following inventways: cleanerswitching materialsto andlow-carbon chemicals,fuel design for fewer materialssources<ref>{{Cite journal|last=KaziBerral|first=RubaiatJosep HabibLl.|last2=GrossmanGoiri|first2=ToviÍñigo|last3=CheongNou|first3=HyunminRamón|last4=HashemiJulià|first4=AliFerran|last5=FitzmauriceGuitart|first5=GeorgeJordi|last6=Gavaldà|first6=Ricard|last7=Torres|first7=Jordi|date=2017-10-202010|title=DreamSketch:Towards Earlyenergy-aware stagescheduling 3Din designdata explorationscenters withusing sketchingmachine and generative design.learning|url=http://dx.doi.org/10.1145/31265941791314.31266621791349|journal=Proceedings of the 30th1st AnnualInternational ACM SymposiumConference on UserEnergy-Efficient Interface SoftwareComputing and TechnologyNetworking - e-Energy '10|location=New York, NYNew York, USA|publisher=ACM|volume=|pages= Press|doi=10.1145/31265941791314.31266621791349|isbn=978-1-4503-49810042-9|via=1}}</ref>, improve production quality, predict machine breakdowns, optimize heating and cooling systems<ref name=":0">Richard Evans and Jim Gao. DeepMind AI reduces Google data centre cooling bill by 40%. DeepMind blog, 20, 2016.</ref>, and prioritize the use of clean electricity over fossil fuels<ref>{{Cite journal|last=Zhang|first=Xiao|last2=Hug|first2=Gabriela|last3=Kolter|first3=J. Zico|last4=Harjunkoski|first4=Iiro|date=2016-07|title=Model predictive control of industrial loads and energy storage for demand response|url=http://dx.doi.org/10.1109/pesgm.2016.7741228|journal=2016 IEEE Power and Energy Society General Meeting (PESGM)|publisher=IEEE|doi=10.1109/pesgm.2016.7741228|isbn=978-1-5090-4168-8}}</ref>; optimizing heating, ventilation, and air conditioning (HVAC) systems<ref name=":0">Richard Evans and Jim Gao. DeepMind AI reduces Google data centre cooling bill by 40%. DeepMind blog, 20, 2016.</ref>; improving production quality and efficiency; predicting equipment failures ahead of time; facilitating discoveries in material science and chemical engineering; designing for more efficient material use<ref>{{Cite journal|last=BerralKazi|first=JosepRubaiat Ll.Habib|last2=GoiriGrossman|first2=ÍñigoTovi|last3=NouCheong|first3=RamónHyunmin|last4=JuliàHashemi|first4=FerranAli|last5=GuitartFitzmaurice|first5=Jordi|last6=Gavaldà|first6=Ricard|last7=Torres|first7=JordiGeorge|date=20102017-10-20|title=TowardsDreamSketch: energy-awareEarly schedulingstage in3D datadesign centersexplorations usingwith machinesketching learningand generative design.|url=http://dx.doi.org/10.1145/17913143126594.17913493126662|journal=Proceedings of the 1st30th InternationalAnnual ConferenceACM Symposium on Energy-EfficientUser ComputingInterface Software and Networking - e-Energy '10Technology|location=New York, New YorkNY, USA|publisher=ACM Press|volume=|pages=|doi=10.1145/17913143126594.17913493126662|isbn=978-1-4503-00424981-19|via=}}</ref>; and streamlining supply chains.
YetNonetheless, greaterimproving efficiencyindustrial mayemissions increaserequires the productionavailability ofand goodssharing andof thushigh-quality GHG emissionsdata ( viawhich theis Jevonsoften paradoxproprietary to firms) , unlessas industrialwell actorsas havethe sufficientalignment of firms' incentives towith reduceactually overallreducing emissions instead of simply producing more goods at a lower cost.<ref>{{Cite journal|last=Sorrell|first=Steve|date=2009-04|title=Jevons’ Paradox revisited: The evidence for backfire from improved energy efficiency|url=http://dx.doi.org/10.1016/j.enpol.2008.12.003|journal=Energy Policy|volume=37|issue=4|pages=1456–1469|doi=10.1016/j.enpol.2008.12.003|issn=0301-4215}}</ref> In addition, ML is most useful when factory processes or complex supply chains can actually be adjusted accordingly, and when objective functions are clear. ▼
ML can have a substantial positive climate impact on the industrial sector under the following circumstances:
* When there is enough accessible, high-quality data around specific processes or transport routes.
* When firms have an incentive to share their proprietary data and/or algorithms with researchers and other firms.
* When aspects of production or shipping can be readily fine-tuned or adjusted, and there are clear objective functions.
* When firms’ incentives align with reducing emissions (for example, through efficiency gains, regulatory compliance, or high GHG prices).
▲Yet greater efficiency may increase the production of goods and thus GHG emissions (via the Jevons paradox) unless industrial actors have sufficient incentives to reduce overall emissions<ref>{{Cite journal|last=Sorrell|first=Steve|date=2009-04|title=Jevons’ Paradox revisited: The evidence for backfire from improved energy efficiency|url=http://dx.doi.org/10.1016/j.enpol.2008.12.003|journal=Energy Policy|volume=37|issue=4|pages=1456–1469|doi=10.1016/j.enpol.2008.12.003|issn=0301-4215}}</ref>.
== Machine Learning Application Areas ==
== Background Readings ==
* '''The IPCC's Climate Change 2014 section on Industry'''<ref name=":1" />''':''' A comprehensive summary of the sources of industrial emissions worldwide and their potential for mitigation. Available [https://www.ipcc.ch/site/assets/uploads/2018/02/ipcc_wg3_ar5_chapter10.pdf here].
* '''Industry 4.0:''' Wikipedia's article on the nebulous concept of Industry 4.0. Available [https://en.wikipedia.org/wiki/Industry_4.0 here].
* '''Life Cycle Assessment (2004)'''<ref>{{Cite journal|last=Rebitzer|first=G.|last2=Ekvall|first2=T.|last3=Frischknecht|first3=R.|last4=Hunkeler|first4=D.|last5=Norris|first5=G.|last6=Rydberg|first6=T.|last7=Schmidt|first7=W. -P.|last8=Suh|first8=S.|last9=Weidema|first9=B. P.|last10=Pennington|first10=D. W.|date=2004-07-01|title=Life cycle assessment: Part 1: Framework, goal and scope definition, inventory analysis, and applications|url=http://www.sciencedirect.com/science/article/pii/S0160412003002459|journal=Environment International|language=en|volume=30|issue=5|pages=701–720|doi=10.1016/j.envint.2003.11.005|issn=0160-4120}}</ref>: A friendly paper introducing the concept of measuring carbon emissions along the entire supply chain and life-cycle of physical products. Available [https://www.sciencedirect.com/science/article/pii/S0160412003002459 here].
|