kth.sePublications KTH
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Advancing an already high-performance smart building with model predictive control: Multi-layer optimization under forecast uncertainty in a real building case
KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Building Technology and Design.ORCID iD: 0000-0002-8118-8329
KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Building Technology and Design.
KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Sustainable Buildings. KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Building Technology and Design.ORCID iD: 0000-0001-5902-2886
Mälardalen University.
Show others and affiliations
2026 (English)In: Applied Energy, ISSN 0306-2619, E-ISSN 1872-9118, Vol. 402, article id 126999Article in journal (Other academic) Published
Abstract [en]

Thermal energy systems in buildings play a central role in global decarbonization efforts, accounting for a significant share of energy use and carbon emissions. This study addresses a key research question: how can advanced control strategies further enhance the performance of already energy-efficient, low-exergy thermal systems in low-energy buildings? To address this, a model predictive control (MPC) framework is designed to optimize the operation of an advanced thermal system based on modern concepts of low-temperature heating and high-temperature cooling, including ground-source heat pumps, borehole thermal storage, and modern air handling units. This approach employs a multi-layered MPC cost function, considering both immediate operational costs (electricity and heating) as well as system impact penalties, such as CO₂ emissions, thermal energy storage preservation, comfort violations, and peak load shaving, in response to fluctuating market cost signals, outdoor temperature, and thermal storage limitations. Applied to a validated, ultra-efficient commercial building, the MPC framework achieves a 13 % reduction in annual market-responsive operational costs, a 20 % improvement in long-term savings, and a four-year shorter payback period compared to existing well-established rule-based control. The results further confirm the robustness of predictive control under realistic forecast errors, as demonstrated by Monte Carlo simulations. From an environmental perspective, the CO₂ emission index stays below both Swedish electricity and district heating baselines, demonstrating the environmental benefits of predictive control through strategic sector coupling. Beyond the case study, the proposed method provides a scalable pathway for integrating predictive control into next-generation smart buildings. It highlights the potential of MPC as the final optimization layer in advanced thermal systems, aligning with global objectives for cost-promising and carbon-neutral building operations. 

Place, publisher, year, edition, pages
Elsevier BV , 2026. Vol. 402, article id 126999
Keywords [en]
Model predictive control (MPC), Forecast uncertainty, Ground source heat pump, Cost penalty optimization, Smart HVAC, Boreholes.
National Category
Energy Systems Energy Engineering
Research subject
Civil and Architectural Engineering, Fluid and Climate Theory; Energy Technology
Identifiers
URN: urn:nbn:se:kth:diva-368046DOI: 10.1016/j.apenergy.2025.126999ISI: 001614844400007Scopus ID: 2-s2.0-105020918060OAI: oai:DiVA.org:kth-368046DiVA, id: diva2:1986743
Funder
Swedish Energy Agency, 51490-1
Note

QC 20250806

Available from: 2025-08-03 Created: 2025-08-03 Last updated: 2025-12-30Bibliographically approved
In thesis
1. Smart Design, Control, and Optimization of Thermal Energy Storage in Low-Temperature Heating and High-Temperature Cooling Systems
Open this publication in new window or tab >>Smart Design, Control, and Optimization of Thermal Energy Storage in Low-Temperature Heating and High-Temperature Cooling Systems
2025 (English)Doctoral thesis, comprehensive summary (Other academic)
Alternative title[sv]
Smart utformning, styrning och optimering av termisk energilagring i lågtemperaturvärme- och högtemperaturkylsystem
Abstract [en]

Heating and cooling account for about 55% of energy used in buildings worldwide and are a leading source of operational CO2 emissions. In cold‑climate regions such as Sweden, supplying this thermal demand efficiently and sustainably is crucial for meeting national and global climate goals. Low‑Temperature Heating (LTH) and High‑Temperature Cooling (HTC) systems, with minimal temperature difference between energy supply and demand, are modern solutions that work well with low‑exergy and renewable energy sources. When paired with Thermal Energy Storage (TES), especially deep seasonal boreholes, these technologies can boost energy flexibility, lower carbon intensity, and improve long‑term system resilience. Yet an important question remains: How can we push these already ultra‑efficient LTH‑HTC‑TES systems even further? The answer lies in adding intelligent control and optimization, the final layer needed to unlock their full operational, economic, and environmental potential.

This thesis investigates how smart design, advanced control, and optimization can make an already smart LTH‑HTC‑TES design even smarter. The study centers on Juvelen, a 10,000 m² commercial building in the city of Uppsala in Sweden, famous for its “deep‑green” thermal energy concept: it drills deep into the ground to utilize the ground's heating and cooling potential directly through borehole TES without any extra machinery while also using passive cooling, energy‑recovery ventilation, and smart interaction with the district‑heating network. This makes Juvelen a perfect testbed for exploring how AI‑driven optimization and control can further improve next‑generation thermal systems in commercial buildings. The research is carried out in three successive stages. First, a comprehensive literature review identifies the knowledge gaps in how TES is integrated, controlled, and optimized within LTH and HTC systems. Second, a detailed dynamic model of Juvelen and its thermal system is developed in TRNSYS and validated against real‑time data from the TEKLA building‑management system. Third, this validated model is used to investigate several smart performance‑enhancement approaches: (i) multi‑objective optimization with an artificial neural network (ANN) surrogate and evolutionary algorithms such as Grey Wolf, Non-dominated Sorting Genetic Algorithm‑II, particle swarm, and dragonfly, (ii) modifying the radiator temperature through an optimal adaptive radiator control fine‑tuned through particle swarm, (iii) evaluation of an alternative system configuration integrated with a ground‑source heat pump, and (iv) a forecast‑driven model predictive control (MPC) tested under realistic uncertainty.

The baseline study shows that, even without mechanical chillers or heat pumps, the existing system meets the entire cooling load and about one‑third of the heating demand directly through the borehole thermal storage, with the remaining heat supplied by the district heating network. Over its ten‑year payback period, the system avoids more than 140 tonnes of CO2, making it an ideal candidate for exploring how even the best-performing systems can be enhanced through smart design, predictive control, and optimization. Building on this foundation, the continuation of the research demonstrates how intelligent methods, and smart configurations can unlock further gains. Among several neural‑network training options, the Levenberg–Marquardt algorithm combined with a cascade‑forward architecture produced the smallest prediction errors and shortest computation times, making it the most dependable for optimization. Using this model, ANN‑based optimization lowered the levelized cost of thermal energy by 14.5%, reduced CO₂ intensity by 6%, and increased delivered thermal energy by 11%. An optimal adaptive radiator‑supply control, derived from a revised temperature equation considering solar radiation, ventilation status, and internal gains, improved comfort by 72.7% on the weighted temperature deviation scale and cut heating bought from the network by about 13.2%. Further benefits came from adding two 40 kW heat pumps, which increased seasonal flexibility. A new seasonal heat‑management scheme lets the district heating network recharge the boreholes each September, keeping the ground in thermal balance and boosting long‑term sustainability. This integration lowered annual operating costs by 9.4% and CO₂ impact from 23.9 tonnes in the existing system to 1.6 tonnes, aligning the proposed smart integration with Sweden’s net-zero building goals. Finally, a forecast‑driven model‑predictive controller achieved more than 13% and 5% extra market-responsive and operational cost savings and shortened payback by about four years, while Monte Carlo tests confirmed its robustness against forecasting errors.

Abstract [sv]

Uppvärmning och kylning står för cirka 55% av den totala energianvändningen i byggnader globalt och är en av de främsta källorna till driftsrelaterade koldioxidutsläpp. I kallare klimatzoner, såsom Sverige, är det avgörande att tillgodose detta termiska behov på ett effektivt och hållbart sätt för att uppfylla både nationella och globala klimatmål. Lågtempererad uppvärmning (LTH) och högtempererad kylning (HTC), där temperaturskillnaden mellan energitillförsel och -behov är minimal, är moderna lösningar som passar väl ihop med lågexergi- och förnybara energikällor. När dessa kombineras med termisk energilagring (TES), särskilt djupa säsongslagrande borrhål, kan systemen erbjuda ökad flexibilitet, minskad koldioxidintensitet och förbättrad långsiktig driftsäkerhet. Trots detta kvarstår en viktig fråga i litteraturen: Hur kan redan ultraeffektiva LTH–HTC–TES-system förbättras ytterligare? Svaret ligger i att addera intelligent styrning och optimering – det sista skiktet som krävs för att frigöra deras fulla operativa, ekonomiska och miljömässiga potential.

Denna avhandling undersöker hur smart design, avancerad styrning och optimering kan göra ett redan intelligent LTH–HTC–TES-system ännu smartare. Studien fokuserar på Juvelen, en 10 000 m² stor kommersiell byggnad i Uppsala, känd för sitt "deep-green"-koncept där borrhål används direkt för att täcka värme- och kylbehovet utan värmepumpar eller kylmaskiner. Byggnaden är utrustad med passiv kylning, värmeåtervinning i ventilationssystemet och smart koppling till fjärrvärmenätet – vilket gör Juvelen till en idealisk testbädd för att undersöka hur AI-baserad styrning och optimering kan förstärka nästa generations termiska system i kommersiella byggnader. Forskningen bedrivs i tre steg: Först identifieras kunskapsluckor i litteraturen kring hur TES integreras, styrs och optimeras i LTH– och HTC-system. Därefter utvecklas en detaljerad dynamisk modell av Juvelens energisystem i TRNSYS och valideras mot realtidsdata från byggnadens styrsystem (TEKLA). Slutligen används den validerade modellen för att undersöka flera prestandaförbättrande metoder: (i) multiobjektiv optimering med artificiellt neuralt nätverk (ANN) och metaheuristiska algoritmer som Grey Wolf, NSGA-II, particle swarm och dragonfly, (ii) en adaptiv styrning av radiatortemperatur baserad på optimerad temperaturformel via PSO, (iii) utvärdering av ett alternativt system med integrerad bergvärmepump, och (iv) modellprediktiv styrning (MPC) baserad på prognoser och testad under osäkerhet.

Grundstudien visar att systemet, utan värmepumpar eller kylaggregat, täcker hela byggnadens kylbehov och cirka en tredjedel av värmebehovet direkt via borrhålen. Resten täcks av fjärrvärme. Under en återbetalningstid på tio år undviks mer än 140 ton CO₂, vilket gör systemet till en utmärkt kandidat för att undersöka hur redan högpresterande lösningar kan förbättras ytterligare genom smart design, prediktiv styrning och optimering. Byggt på denna grund visar fortsatta undersökningar hur intelligenta metoder och smarta konfigurationer ger ytterligare vinster. Levenberg–Marquardt-algoritmen i kombination med en cascade-forward-arkitektur visade lägst fel och snabbast beräkningstid bland flera ANN-träningsmetoder. Med denna modell kunde ANN-baserad optimering minska energikostnaden med 14,5 %, minska CO₂-intensiteten med 6 % och öka levererad termisk energi med 11 %. En nyutvecklad adaptiv styrstrategi för radiatorsystem – som tar hänsyn till solinstrålning, ventilation och interna värmelaster – förbättrade komforten med 72,7 % (baserat på viktad temperaturavvikelse) och minskade köpt värme från nätet med cirka 13,2 %. Ytterligare förbättringar uppnåddes genom att integrera två 40 kW värmepumpar, vilket ökade den säsongsvisa flexibiliteten. Ett nytt styrkoncept introducerades där fjärrvärmenätet laddar borrhålen varje september, vilket bibehåller termisk balans i marken och stärker långsiktig hållbarhet. Denna integrering sänkte de årliga driftskostnaderna med 9,4 % och CO₂-utsläppen från 23,9 ton till 1,6 ton per år, i linje med Sveriges klimatmål för nettonollbyggnader. Slutligen uppnådde den prognosbaserade MPC-styrningen mer än 13 % kostnadsbesparing på marknadsresponsiv drift och 5 % total driftskostnadsreduktion. Återbetalningstiden minskade med ungefär fyra år, samtidigt som robustheten mot prognosfel verifierades genom Monte Carlo-simuleringar.

Place, publisher, year, edition, pages
Stockholm: KTH Royal Institute of Technology, 2025. p. 105
Series
TRITA-ABE-DLT ; 2520
Keywords
Low-Temperature Heating, High-Temperature Cooling, Smart TES, AI-Driven Multi-Objective Optimization, Adaptive Control, Smart GSHP, Model Predictive Control, Lågtempererad uppvärmning, Högtempererad kylning, Smart TES, AI-baserad multiobjektiv optimering, Adaptiv styrning, Smart GSHP, Modellprediktiv styrning
National Category
Building Technologies Energy Systems Energy Engineering
Research subject
Civil and Architectural Engineering, Fluid and Climate Theory
Identifiers
urn:nbn:se:kth:diva-368012 (URN)978-91-8106-364-6 (ISBN)
Public defence
2025-08-28, Kollegiesalen, Brinellvägen 8, KTH Campus, Public video conference link [MISSING], Stockholm, 14:00 (English)
Opponent
Supervisors
Funder
Swedish Energy Agency, 51490-1
Note

QC 20250805

Available from: 2025-08-05 Created: 2025-08-01 Last updated: 2025-12-17Bibliographically approved

Open Access in DiVA

fulltext(7216 kB)29 downloads
File information
File name FULLTEXT01.pdfFile size 7216 kBChecksum SHA-512
8ef4ca69cddb9997e99e928a0ef27ff7986df64badf9015a8aa3f99aa84e1237b8c83ca238df5dddef4447a0d9bf6d28ad55a08b7b189a4bc1dad0fb36789c01
Type fulltextMimetype application/pdf

Other links

Publisher's full textScopus

Authority records

Behzadi, AmirmohammadPloskic, AdnanSadrizadeh, Sasan

Search in DiVA

By author/editor
Behzadi, AmirmohammadGoudarzi, NaserPloskic, AdnanSadrizadeh, Sasan
By organisation
Building Technology and DesignSustainable Buildings
In the same journal
Applied Energy
Energy SystemsEnergy Engineering

Search outside of DiVA

GoogleGoogle Scholar
Total: 29 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

doi
urn-nbn

Altmetric score

doi
urn-nbn
Total: 186 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf