Conversion to Fourth-Generation District Heating (4GDH): Heat Accumulation Within Building Envelopes

Abstract

This study investigates improving district heating (DH) systems by analyzing the effects of low-temperature operation on network efficiency, heat losses, and indoor temperature stability. A mathematical model is developed to simulate building heat performance under different supply temperatures, substation connection types, and envelope materials. The methodology involves detailed hourly heat load simulations and optimization techniques to assess the impact of temperature flexibility and heat accumulation within buildings. The results reveal that a 10 °C reduction in supply temperature leads to a heat loss decrease of up to 20%, significantly improving system efficiency. Moreover, buildings with higher thermal inertia and indirect substation connections exhibit better resilience to short-term temperature fluctuations, ensuring more stable indoor conditions. The analysis also demonstrates that optimizing temperature control can reduce operational costs by 19%, primarily by minimizing excessive heat supply and utilizing stored thermal energy effectively. Despite slight temperature fluctuations in extreme conditions, the system maintains indoor comfort levels within acceptable limits. This study concludes that transitioning to a lower-temperature DH system is feasible without compromising reliability, provided heat accumulation effects and supply flexibility are carefully managed. These findings offer a replicable approach for improving DH efficiency in networks with diverse building configurations.

1. Introduction

1.1. Motivation and the Research Gap

The escalating global energy demand, pressing sustainability goals, and the urgent need for climate change mitigation underscore the importance of energy efficiency across all sectors. As the world’s population grows and economies develop, the strain on finite energy resources intensifies, leading to environmental degradation, geopolitical instability, and economic vulnerabilities. Improving energy efficiency—achieving more output with less energy input—offers a crucial pathway for decoupling economic growth from increased energy consumption and its associated negative impacts. Reducing the overall energy demand can alleviate the pressure on energy production infrastructure, conserve natural resources, decrease greenhouse gas emissions, and enhance energy security. This makes energy efficiency a cornerstone of any strategy for achieving a sustainable and resilient energy future.

The building sector is a particularly significant area for energy efficiency improvements. Globally, buildings account for a substantial portion of total energy consumption and associated carbon emissions. Heating, ventilation, and air conditioning (HVAC) and domestic hot water (DHW) systems represent major energy end-uses within buildings. District heating (DH) systems play a vital role in supplying thermal energy to urban areas, offering potential advantages in terms of fuel flexibility, economies of scale, and reduced local emissions compared to individual building-level heating solutions. However, inefficiencies within DH networks, such as heat losses during distribution and suboptimal operational strategies, can significantly diminish their environmental and economic benefits. Therefore, enhancing the energy performance of buildings connected to DH systems and optimizing the operation of the DH networks themselves is crucial for realizing the full potential of district heating in a sustainable energy landscape.

Despite the recognized importance of energy efficiency in DH systems and connected buildings, a notable gap exists in the comprehensive understanding and effective implementation of strategies that holistically integrate building thermal characteristics with DH network operation. While considerable research has focused on optimizing individual components of the system—such as improving insulation materials, enhancing boiler efficiency, or developing advanced control algorithms for network flow—less attention has been paid to the dynamic interplay between the thermal behavior of diverse building types and the operational parameters of the DH network, particularly under varying conditions and in the context of advanced operational strategies like low-temperature district heating (LTDH). The existing literature often treats buildings as uniform loads or focuses on specific building typologies in isolation, neglecting the heterogeneity of the building stock connected to a typical DH network and the implications of this diversity for overall system efficiency and stability [1,2]. Furthermore, the potential for utilizing the thermal inertia of building envelopes as an inherent form of distributed energy storage within the DH system remains largely underexplored in conjunction with flexible network operation. More details are listed in Section 1.2.

This study addresses this gap by investigating the impact of building thermal performance characteristics, specifically focusing on the heat capacity and heat loss rates of different building typologies connected to a DH network. By analyzing the thermal response of various building types to fluctuations in DH supply temperature and external weather conditions, we seek to quantify the potential for leveraging building thermal inertia to enhance the efficiency and flexibility of DH system operation. Specifically, this research focuses on evaluating the benefits of LTDH in conjunction with the inherent heat accumulation capacity of building envelopes, aiming to demonstrate how a more integrated approach, considering both the DH network and the connected buildings as a coupled thermal system, can lead to significant reductions in heat losses, improved network stability, and enhanced overall energy efficiency. The analysis presented in the subsequent sections utilizes real-world case studies and simulation data to provide concrete insights into the potential of this integrated approach.

1.2. Previous Works

Meesenburg et al. [3] compare several low-temperature options, considering an SH share and involving a heat pump in a generation. They conclude that the feasibility of a configuration depends on the plot ratio of the supplied area and the proportion between capital costs in a DH plant and the additional investments in local facilities. Through simulation, it was observed that the developed control mechanism successfully stabilized the system’s temperature, keeping it consistently within a range of 2 Kelvin around the pre-defined value (like the temperature of a room). Buffa et al. [4] simulate the coordinated balancing effects of several local units. This idea is more beneficial than the free-floating temperature approach in unidirectional and bidirectional DH systems. Low-temperature DH networks can be implemented without the obligate integration of heat pump technology [5].

On the other hand, the essential element of any DH system is an energy transfer station—a substation. As in [6], this study evaluates various substation models to determine their suitability for characterizing network flexibility within DH systems. In Vandermeulen et al.’s papers [6], temperature flexibility is achieved by temporarily increasing the supply temperature. An accurate model of a DH system behavior is essential to evaluate the benefits obtained. Based on such a model, Jebamalai et al. [7] conclude which level of heat storage is best for hourly usage and which for seasonal usage. The options are accumulating heat into building envelopes, utilizing either centralized storage or substation thermal energy accumulators. However, Jebamalai et al.’s [7] focus was not on the flexibility itself but on the detail of the influence of these combinations mentioned above.

Heat demand profiles show that storage systems have smoother consumption and lower peaks than instantaneous DHW substations [8,9]. Kristensen et al. [10] use a Bayesian model to study their influence, link building, and archetype levels, factoring in building diversity. Superposing multiple curves of DHW consumption of instantaneous systems also results in more evenly distributed heat demand profiles with realistic behavior [8,11]. Kristensen et al. [10] leverage the previously defined archetype segmentation to calibrate uncertain input parameters in physics-based building energy models.

Other scientists have faced the same obstacles since the historical supply temperature. For example, according to the operational data, the DH system in the Skultuna area, Sweden, can achieve temperatures up to 110 °C. Saletti et al. [12] successfully cover another focus of their research—reducing heat losses. That makes our studies toward low-temperature DH closer and simpler to compare.

1.3. Use Case and Practical Applications

A literature review indicates that the feasibility of a concept can be assessed by comparing annual heat and electricity generation, taking into account the network section’s return temperature, as demonstrated in [13]. The DH service users might also sense a positive effect, although indirectly, as the potential for heat price reduction through operator gains results in a benefit shared by all district heating consumers. Like us, Barone et al. [14] chose the week from 1–7 January 2020, to thoroughly study time histories of supply and return temperatures for four sample substations in Naples, Italy.

This work introduces a novel approach by analyzing the transition from low-temperature to standard (150/70) operation during increased heat demand periods, such as winter’s onset. Unlike Hammer et al. [15], who suggest shutting down the DH network during low-demand periods with high heat losses and relying on decentralized thermal storage, our method considers pre-charging building envelopes before full-load operation. This pre-charging strategy results in a significantly reduced indoor temperature drop compared to a standard scenario.

The impact of building types and substation connections on energy performance is another area that requires further investigation. Although previous studies categorize buildings based on their construction year, substation connections, and envelope types, a deeper understanding of how these factors influence energy efficiency and heating demand over extended periods is essential. This is particularly important in large-scale urban settings, where building diversity can significantly affect system performance. Moreover, while this study acknowledges the challenges posed by older building stock, more data on the energy performance of these buildings, especially in comparison to newer, more energy-efficient structures, would be valuable. Understanding how DH systems can be upgraded to accommodate older buildings is crucial for developing strategies to reduce energy consumption in urban infrastructures.

1.4. Objectives, Novelty, and the Outline

The contributions of this study lie in its methodology, which provides insights into the feasibility of transitioning district heating systems toward low-temperature concepts. It demonstrates the potential of low-temperature and flexible temperature operation for reducing energy consumption and operational costs in DH systems, particularly by incorporating heat accumulation in building envelopes. This study also offers a detailed case study of Omsk, providing valuable data on how different building types and substation connections affect heating demand and system performance. This work serves as a model for assessing other district heating systems, offering a basis for further research and optimization in similar climates and urban settings. Additionally, this study emphasizes the importance of system flexibility and efficient temperature management, which could lead to significant energy savings and improved system stability in the future.

The mathematical model developed in this study introduces several innovative aspects that differentiate it from existing models used in DH system analysis. Unlike conventional models that often rely on static or annual average values, this model employs dynamic optimization using mixed-integer linear programming with an hourly resolution. This level of granularity allows for a more precise assessment of heat demand fluctuations and the potential for temperature flexibility. Additionally, the model integrates heat accumulation within building envelopes, a factor often neglected in previous studies, to assess its impact on supply temperature regulation and overall system efficiency.

One key innovation is incorporating flexible temperature control, considering immediate and delayed responses to outdoor temperature variations. This feature enables a better evaluation of how supply temperature reductions influence indoor thermal stability, energy consumption, and system-wide heat losses. The model also accounts for real operational constraints, including regulating valve dynamics and the influence of substation connection types on secondary loop temperature variations. Furthermore, it provides an in-depth analysis of how different envelope materials contribute to heat retention and dissipation, offering valuable insights into optimizing DH network performance in diverse building environments.

The introduction sets the context by outlining this study’s objectives. Following this, the methodology section details the approach, including analyzing different building types, substation connections, envelope materials, and the operational data used to assess the heating demand. The case study describes particular buildings connected to a DH system in Omsk. The results and discussion section presents the findings from the simulations, focusing on heat demand variations, energy consumption, and the impact of different system configurations. This section also compares the results with other studies and discusses the potential benefits of low-temperature operations in DH systems. This paper concludes with a summary of the key findings, contributions to the field, and recommendations for future research, particularly in relation to integrating low-temperature heating and optimizing DH systems in similar climates.

2. Materials and Methods

This study employs a structured methodology to analyze the heating performance of different building types within the DH system. The approach integrates data collection, modeling, and simulation to assess the impact of LTDH operation on energy efficiency and heat loss reduction. The methodology consists of the following five components:

• Selection of case study and building types;
• Data collection and heat demand analysis;
• Mathematical modeling and simulation;
• Assessment of low-temperature operation and system flexibility;
• Validation and comparative analysis.

To capture the diversity of buildings connected to the system, representative building cases were selected based on three primary characteristics: substation connection type (either direct or indirect connection to the DH network), building type (office buildings and residential buildings), envelope material (brick, prefabricated panels, or foam concrete wall blocks). This categorization results in nine unique building scenarios, comprehensively comparing different heating configurations and their energy performance.

This study analyzes heating demand based on real operational data collected between 11 November and 17 November, a transition period between moderate and cold weather. The primary focus is SH demand, with data from previous research and official records on building energy consumption. The analysis considers outdoor temperature profiles (to capture seasonal variations affecting heat demand), heat accumulation ability (evaluated based on building materials and construction details, supported by the previous literature), and supply and return temperature trends (observed to understand system behavior during cold periods).

These parameters help determine different DH configurations’ energy efficiency and potential transition toward LTDH operation.

This study employs a mathematical model to simulate heat transfer and energy demand under different operational scenarios. The model incorporates building envelope characteristics (thermal conductivity, glazing ratio, and specific heat loss coefficients for each building type) and heat transfer equations to estimate heat accumulation, losses, and energy consumption over time. The mathematical model was compared with previous studies that used annual-resolution data, ensuring a more dynamic and realistic assessment.

To evaluate the feasibility of LTDH, different temperature scenarios were tested: conventional high-temperature DH operation (supply/return temperatures of 95/70 °C or higher), LTDH (reducing the supply temperature to 50–60 °C while maintaining efficient heat delivery), flexible temperature operation (dynamic supply temperature adjustments based on outdoor conditions and heat accumulation properties).

Indoor temperature profiles, heat loss reduction, and energy savings were analyzed for different cases. This study also investigated the impact of flexible temperature operation on compensation times and system stability.

The results were validated against reference buildings from similar studies, ensuring the model’s accuracy in estimating heat demand and losses. The findings were further compared with case studies from other DH networks, such as those in Estonia and Sweden, to highlight similarities and differences in system performance under varying climatic conditions.

Like Turski et al. [16], this approach utilizes the specific heat of water to accumulate energy by artificially raising the supply flow temperature. The difference is the model we apply. The model was adopted as per the Russian scientist Koryagin M.V. (unpublished):

$$\delta t_{in}=\mu \cdot {\displaystyle \frac{Q_{\max }}{F_0}}\cdot \left({\displaystyle \frac{{\Omega }_{\max }-{\Omega }_{\min }}{Y_0}}+{\displaystyle \frac{{\eta }_{om}}{\overline{{\alpha }_{к}}}}\right)\cdot \xi ,$$

(1)

$$\Delta t_{in}^{-}=\mu \cdot {\displaystyle \frac{Q_{\max }}{F_0}}\cdot \left({\displaystyle \frac{{\eta }_{om}}{\overline{{\alpha }_{к}}}}-{\displaystyle \frac{{\Omega }_{\min }}{Y_0}}+{\displaystyle \frac{F_{0}m}{q_0}}\right)\cdot \xi ,$$

(2)

$$\Delta t_{in}^{+}=\delta t_{in}-\Delta t_{in}^{-},$$

(3)

where μ is a factor to compensate for error due to replacing a vector with a scalar;

Q_max is the building heat demand [W];

F₀ is the area of the envelopes’ surface [m²];

Y₀ is the overall heat transfer coefficient [W/(m²∙°C)];

${\displaystyle \frac{{\eta }_{om}}{\overline{{\alpha }_{к}}}}$ is the factor of combined convection and thermal radiation heat transfer;

ξ is a factor that characterizes the ability of furniture to absorb heat;

Ω is a factor indicating that solar and internal heat gains are non-steady-state;

q₀ represents building heat losses [W/°C].

The discharging time is equal to the time, which is enough for the indoor temperature drop ν_in(z), adopted from the ideas of Russian scientist Koryagin M.V. (unpublished):

$${\nu }_{in}\left(z\right)={\displaystyle \frac{{{\nu }^{\prime }}_{01}}{1-\beta \cdot m}}\cdot e^{-z\cdot m}+\left({{\nu }^{\prime }}_{01}-{\displaystyle \frac{{{\nu }^{\prime }}_{01}}{1-\beta \cdot m}}\right)\cdot e^{-z/\beta },$$

(4)

If charging is a fast process, which may take place when there is negligible heat capacity of an SH system available, i.e., if a substation is connected directly (with no heat exchanger and no hydraulic separation of the primary and secondary circuits),

ν_in(z) = ν₀₁′∙e^−z/β,

(5)

and when the forecasted outdoor temperature t_out drops below the current temperature t_out.c,

t_out = t_out.c + ΣQ_ihg/ΣQ_sp.hl = t_out.c + ν_t,

(6)

where ΣQ_ihg is internal heat gains [W];

ΣQ_sp.hl is specific heat losses [W].

The building’s heat accumulation capacity allows for substantial compensation periods, enabling the balancing of heat source output. Thus, z and β values can be considered to analyze the time and thermal performance of building envelopes to withstand temperature drops to the lowest values of indoor temperature t_in.l (Equations (7) and (8), respectively):

$$z=\beta \cdot \ln {\displaystyle \frac{k_t\cdot (t_{in}-t_{out}-{\displaystyle \sum Q_{ihg}}/{\displaystyle \sum Q_{sp.hl}})}{t_{in.l}-t_{out}-{\displaystyle \sum Q_{ihg}}/{\displaystyle \sum Q_{sp.hl}}}},$$

(7)

where k_t is a factor indicating the number of floors, glazing ratio, type of SH system, etc.;

t_in is the design indoor temperature [°C];

β [h] is

$$\beta ={\displaystyle \frac{k_t{\displaystyle \sum {\delta }_i{c}_i{\rho }_i{F}_i/2}}{3.6\left[{\displaystyle \sum k_j{F}_j+Lc\rho }\right]}},$$

(8)

where δ_i is the thickness of the i-th insulation layer [m];

ρ_i is the density of the i-th insulation material [kg/m³];

c_i is the specific heat capacity of the i-th insulation material [W/kg°C];

F_i is the surface area of the i-th insulation material [m²];

k_j is the heat transfer coefficient [W/m²°C];

F_j is the surface area of the building envelope [m²];

L is the infiltration rate [m³/h];

ρ is the density of infiltrated air [kg/m³];

c is the specific heat capacity of infiltrated air [W/kg°C].

The cost savings are assessed with the assumptions of the potential of low-temperature operation with no indoor temperature drop, which reduces heat losses during fall and spring, and a heat price of 30 EUR/MWh.

The linear heat loss density for each buried pipe, W/m, is given by

$$q_s^{und}={\displaystyle \frac{\pi ({\tau }_s^{design}-t_{soil}^{design})}{\frac{\ln \frac{d_{i.s}^{und}+2\delta }{d_{i.s}^{und}}}{2\pi \lambda }+\frac{\ln \frac{4h}{d_{i.s}^{und}+2\delta }}{2\pi {\lambda }_{soil}}}},$$

(9)

$$q_r^{und}={\displaystyle \frac{\pi ({\tau }_r^{design}-t_{soil}^{design})}{\frac{\ln \frac{d_{i.r}^{und}+2\delta }{d_{i.r}^{und}}}{2\pi \lambda }+\frac{\ln \frac{4h}{d_{i.r}^{und}+2\delta }}{2\pi {\lambda }_{soil}}}},$$

(10)

where ${\tau }_s^{design}$ and ${\tau }_r^{design}$ represent the design feed and return line temperatures, respectively [°C];

$t_{soil}^{design}$ is the design soil temperature [°C];

$d_{i.s}^{und}$ and $d_{i.r}^{und}$ are the i-th network section’s feed and return diameter, respectively [m];

δ is the insulation thickness [m];

λ is the thermal conductivity of the insulation material, W/(m°C);

h is the pipe depth, m;

λ_soil is the heat conductivity coefficient of the soil at the actual average monthly temperature, W/(m°C).

For the thermal conductivity λ, we adopted the value from Dalla Rosa et al.’s paper [17]. In contrast, Hammer et al. [15] used 0.03 W/(m°C).

Due to the absence of operational data, we applied the soil’s mean heat conductivity coefficient at the measured temperature. Dalla Rosa et al. [18] employed 1.6 W/(m°C).

Unlike this study, Meesenburg et al. [3] estimated energy loss empirically, dividing the total annual heat demand by the estimated trench length. That calculation incorporated annual specific heat demand, the plot ratio, and the characteristic width. Arabkoohsar et al. [5] assessed energy losses as a derivative of dimensionless parameters based on line temperatures and diameters. Cai et al. [19] calculated temperature drops along pipes using soil temperature and heat transfer coefficients, but not heat losses directly. However, like Ref. [3], they adjusted heat losses based on supply and return temperatures.

Energy losses for the aboveground pipes, W/m, are given by

$$q_s^{above}={\displaystyle \frac{\pi ({\tau }_s-t_{out})}{\frac{\ln \frac{d_{i.s}^{above}+2\delta }{d_{i.s}^{above}}}{2\pi \lambda }+\frac{1}{2\pi \alpha \left(d_{i.s}^{above}+2\delta \right)}}},$$

(11)

$$q_r^{above}={\displaystyle \frac{\pi ({\tau }_r-t_{out})}{\frac{\ln \frac{d_{i.r}^{above}+2\delta }{d_{i.r}^{above}}}{2\pi \lambda }+\frac{1}{2\pi \alpha \left(d_{i.r}^{above}+2\delta \right)}}},$$

(12)

where α is the heat transfer coefficient at the ground surface, including convection and radiation ranging from 11.6 (if the wind is negligible) to 29 W/(m²°C) (the maximum value according to the national design guideline):

$$\begin{array}{c}\alpha =11.6+7\sqrt{\mathrm{w}},\end{array}$$

(13)

where w is the velocity of wind [m/s].

The simulation environment used in this research is based on ZuluThermo© (Saint-Petersburg, Russia, Politerm, LLC) and Microsoft Excel, where all calculations and scenario analyses were implemented using built-in mathematical functions and formulae. This study leverages Excel’s computational capabilities to model the heat demand, supply temperature variations, and thermal inertia of buildings within the DH network of Omsk. The mathematical model incorporates heat transfer equations, energy balance formulations, and optimization conditions, ensuring the simulated results align with real-world operational data.

The equations provided in this study are embedded in the Excel environment, allowing for iterative computations and scenario testing under different supply temperature settings. ZuluThermo© is utilized for assessing the temperature and flow rate adjustments, enabling the analysis of heat accumulation effects and temperature compensation times across various building types and substation configurations. This approach ensures transparency and flexibility, making it possible to fine-tune the input parameters and observe the resulting variations in energy efficiency, heat losses, and indoor temperature stability.

Additionally, MS Excel facilitates statistical evaluations by processing large datasets of temperature profiles and demand curves. The model simulates how different envelope materials, building heights, and substation connection types influence the DH system’s thermal performance by integrating data tables and conditional logic functions. The ability to visualize trends through Excel’s charting tools provides an intuitive way to interpret how supply temperature reductions impact overall network efficiency and operational costs.

3. Case Study

The city of Omsk has had DH since the late 1930s. The number of consumers is approximately 10,000.

This study considered two types of substation connection (direct or indirect), two consumer types (office and residential buildings, e.g., Figure 1), and three envelope materials, i.e., nine combinations in total.

For comparison, in [10], consumers were initially grouped into 11 categories, drawing partly from the classification system devised by the Danish Building Research Institute as part of the EU-funded IEE projects TABULA and EPISCOPE. These projects focused on defining residential building archetypes across 20 European countries. Unlike Kristensen et al. [10], not only the construction year of the buildings was considered for the segmentation.

In this study, building load and demand refer exclusively to consumers’ SH needs, determined based on operational data collected from November 11th to November 17th (Figure 2). This period was chosen as it represents a typical transition phase between moderate and cold weather conditions (Figure 3).

This heat accumulation ability was evaluated based on the factors and typical construction details available in the previous authors’ papers [20,21] and corroborated by the comparison of assessed energy consumption for reference buildings from each consumer type, substation connection, and construction period (e.g., 2005–2010, Figure 4) available within the papers from Na et al. [22] and Gan et al. [23].

Although primary supply and return temperatures have tended to decrease in recent years, because of the considerable share of old buildings for which the high temperature of water entering an SH system is still of high importance, the temperature regime within the system remains high, up to 111/70 °C during the coldest days in winter. Absolutely the same trend was reported by Volkova et al. [13], although they detected maximum values of 120/70 °C. At the same time, we have found out that in spring and autumn, the DH system is already operating at approximately 70 °C in the feed pipe [24]. This study goes beyond mere data analysis, serving as a case-study-driven demonstration of a methodology that can be applied to assess other DH systems. Moreover, it provides insights into whether these systems are prepared to transition toward low-temperature concepts. Nonetheless, to involve renewables in a generation, better feedback might be obtained for the operating conditions of 50–60 °C, which are close to 70 °C in fall and spring but can only be achieved utilizing temperature flexibility and heat accumulation.

Both design scenarios were modeled using hourly profiles in MS Excel with the specialized software ZuluThermo© (Version 10.0.0.8075u). In [25], the authors optimized similar things through mixed-integer linear programming, but yearly resolutions were applied.

Table 1. Design properties affecting heat demand of the buildings in Omsk obtained for the case study.

Building	Construction Year (Period)	Number of Floors	Type of Building	Type of Substation Connection	Case
Envelope type: brick
a	2007	14	Residential	Indirect	#2
b	2000s	5	Office	Direct
c	1953	3	Residential	Direct	#5
Envelope type: foam concrete wall blocks
d	2020	9	Office	Indirect	#4
e	1990s	9	Residential	Direct	#3
f	2000s	9	Office	Direct
Envelope type: prefabricated panel
g	1960s	14	Residential	Indirect	#6
h	1960s	5	Office	Direct
j	1970s	5	Residential	Direct	#1

summarizes the inputs used in this paper.

We also performed a statistical study of the neighborhood, linking building heat performance based on the building construction year and the glazing ratios (Figure 5) based on the randomized demand profiles.

The same results of the statistical survey are presented by Andric et al. [26]. For detailed information on how the number of floors is considered for calculating heat demands, please check the authors’ recent paper [27].

The design heat demand of the considered buildings ranges from 0.4 to 1.1 MW per building. All nine buildings are the same distance from the DH plant; six are picked for a thorough look. All the assessments have been performed considering the return temperature of 40–45 °C for the return line after the properly tuned regulating valve during a warm day, in which the outdoor temperature is above 0 °C, and the fact that the substation typically decreases the supply temperature from 60 to 50 °C during these conditions despite the 95/70 design temperature regime. For comparison, Arabkoohsar et al. [5] study a DH system where the design outdoor temperature is −25 °C and a heat pump boosts the supply temperature from 40 to 50 °C.

4. Results and Discussion

Based on Table 1, the considerations of the building energy performance are listed in

Table 2. C_env, Q_sp.hl, and β-values for the considered cases. Adopted as per the Russian scientist Koryagin M.V. (unpublished).

Case #	Cenv [10−2] [kJ/°C]	Qsp.hl [W/°C]	β [h]
#1	102.6	53.5	53
#2	123.6	29.6	116
#3	113.9	39.1	80.9
#4	99.7	31.8	87
#5	129	52.4	68.4
#6	73.6	29.3	69.7

.

Considering both the type of insulation and the topology of the DH system, applying Equations (9)–(13), we found that the reduction in heat losses for the case study compared to the baseline scenario is up to 20% with a supply temperature drop of 10 °C (detailed below). That is quite a good achievement since around 15% of the heat produced is typically lost [28]. For comparison, heat losses in the Lahekalda network, Estonia, account for 236.1 MWh (6%), with the possibility of further reduction to 205.2 MWh (4%) [13].

Figure 6 presents the evolution of indoor temperature for the reference-case scenarios #1–6 and for excessive heat from the DH plant provided in a potential case of combined low- and flexible-temperature operation.

In this figure, the indoor temperature of the farthest nodes from the substation is close to 5 °C almost immediately after the outdoor temperature drop of 140–141 h. In contrast, the temperatures of the nearly located buildings (e.g., #2, #3, and #4) are still fine, mainly affected by the type of substation connection and envelope material. However, the pertinent point here, in Figure 6, is an overheating effect when the supply temperature is manually set higher than necessary (previously obtained when the heat demand was still moderate—97..109 h); although flexible temperature operation is outside the scope of this research, it is here to highlight the relevance of the present stage of research. To achieve higher temperatures on such an occasion, a training period was necessary; the compensation time was initially 47 h (94..131 h), but now, it is limited to 18 h (145..163 h). Turski et al. [16] determined compensation times, combined with the time resulting from the heat accumulation ability of the DH network, and they were almost the same: 24 and 48 h.

The heat transfer and energy flux [W/m] results are not depicted but show a similar trend to costs and losses—almost a direct correlation with supply temperature. The seasonal network heat losses decreased from 14.1% for the low-temperature option. They reached a minimum of 11.3%, with the associated savings of EUR 20.2 million, saving excess heat during 60–70 °C operation, including heat accumulation.

Figure 7 summarizes the results of a sample week (1–7 January).

From Figure 7, we conclude that indoor temperatures mostly range between 4 and 15 °C (0 °C represents the minimum threshold temperature obtained during the 163rd hour); for others, it was still between 13 and 20 °C, with two exclusions only.

For the simulation data of buildings #2, #3, and #4, the acceptable indoor temperature considering only the SH system pump electricity consumption lasts as long as 25–27 h. Here, the reason is that the volume of hot water circulating in the secondary loop is 25%. Heat losses account for about the same 5% of the supplied thermal energy, as Buffa et al. [4] report, and still, a drop in temperature in the SH system and, consequently, the indoor temperature to 0 °C is inevitable. In #2, heat storage is already operated with short cycle times to buffer peak heat demands. For #3, residual heat demands occur less frequently and are less prominent due to the thermal performance of envelopes and less heat capacity (Q_sp.hl is only 39.1 W/°C). These combinations of material/height/substation types are the best for constructing new office and residential buildings. The most interesting case is #4, where C_env accounts for 99.7·10⁻² kJ/°C, although the glazing ratio is as much as 40%. The situation has also improved with the type of building. An office building with no DHW consumption shows the most stable heat consumption profile. That makes using heat storage much less frequent and leads to larger cycle times. Wirtz et al. [29] detail the same correlation between heat consumption and storage use and conclude that cold storage is operated with the shortest cycle times over the year.

Basically, profiles on an hourly time interval show a much-reduced peak temperature drop. For a modern building and the default-mode heat production, the indoor temperature is reduced only up to 53% (11 °C) due to the higher heat capacity and/or an indirect connection to a substation. The reduction is more visible for buildings #1, #5, and #6, accounting for 0, 20, and 25% of the nominal value at 16 and 23 h intervals, respectively. We also limit indoor temperature drop to 17 °C (#1, if temp flex is applied). For comparison, Turski et al. [16] found that it was promising to increase heat production by 9.6 MW with a compensation time of 19.8 h. Similarly, a decrease in the energy delivery by 4.8 MW with a compensation time of 43.8 h is possible. According to their scenario, the indoor temperature is reduced by 5 K. Braas et al. [8] achieve peak load reduction by 53% at a 3 min time interval.

Based on the data obtained, the conventional approach was statistically assumed to be better for an average monthly outdoor temperature of −8 °C and higher (before the temperature drop during 145–146 h). The trend obtained for the LTDH, highlighted by the regression fit lines added to the graph, is drastically decreased compared to the reference scenario. Moreover, the new supply temperature is kept steadier, particularly when the outdoor temperature is relatively low. That complies well with Saletti et al.’s [12] findings.

Despite low fluctuations, main discharge, and late discharge periods, the model works well for both modes. The problem is that for the rebound and early discharge phases, even a small difference (up to 1 °C) in the primary supply temperature may result in significant daily variations. However, average errors may be negligible [6]. Despite being next to their threshold boundary, both low-temperature and reference scenarios indicated that the indoor temperature had deviated from its design value, as shown for test building #1 (refer to Figure 6 and Figure 7). According to the DH expert, a possible solution is the proper function of a substation control system to prevent short fluctuations.

The control logic formulated in the previous work [30] did not account for the low-temperature concept or the simultaneous operation of other technologies, such as TES or heat accumulation. The substation design temperature was set at 110 °C, with a thermostat adjusted to a DOT of −20 °C, covering 97% of outdoor conditions [30]. Additionally, it was recommended to set the supply temperature to 50 °C during fall and spring when the average five-day outdoor temperature is 0 °C or higher. Even −8 °C is acceptable as a threshold value, as supported by Figure 3, but valve dynamics are much more important. They should never be ignored when considering flexible operations.

When heat accumulation within building envelopes facilitates the transition to LTDH, the supply temperature drops below 60 °C, reducing the average temperature difference between the distribution network and the surrounding soil by 38%. This, in turn, lowers network heat losses. However, this approach is only feasible if outdoor temperatures below −15 °C are rare or do not persist for long periods in spring or fall. Over the past decade, even in winter, temperatures below −20 °C have never lasted for five consecutive days [24], supporting the viability of this concept. For comparison, Saletti et al. [12] reduced the supply temperature to below 80 °C, achieving a 20% reduction in the temperature difference between the distribution network and the soil, leading to lower heat losses. Additionally, their findings showed a 16% reduction in peak loads and a 23% decrease in mass flow rates, contributing to an apparent drop in pumping power. However, to simplify the analysis, the flow rate at the substation is still assumed to remain fixed, as in practice, the regulating devices and valves almost achieve it.

The key finding is that incorporating low-temperature operation with centralized storage reduces O&M costs by 19% compared to a scenario without heat accumulation, specifically at operating temperatures of 50–60 °C. The maximum storage capacity allows for a 16 h delay in transitioning to normal operation (as observed in building #1) when all studied buildings contribute to limiting abrupt flow rate increases, provided the minimum indoor temperature remains above 0 °C to prevent damage to the space heating system. However, such extreme conditions are rarely encountered in practice and serve only as an upper threshold, as supply temperature adjustments typically take up to five hours. Moreover, weather forecasts are generally accurate, enabling proactive measures to be implemented several hours before an outdoor temperature drop. These insights apply similarly to other areas within the DH system. For instance, for the city scale (approximately 3000 MW substation heat demand), the maximum network cost reduction is about 5–6% with an acceptable 10 h delay since most of the residential buildings are directly connected, and there are up to 30% of houses in Omsk older than those constructed in the 1970s. However, most of them are made of brick rather than panel-built and are more like building #5 from the heat accumulation point of view. Nevertheless, that requires a meticulous study and represents the future area of research.

The building-level storage further provides a higher maximum network cost reduction of 6.57%. Still, it is combined with a TES, whose capacity is 1000 m³ and which can store the hot water with negligible heat losses for at least 24 h [7]. Jebamalai et al. [7] also conclude that optimum storage capacity is possible for all cases and depends on storage distribution.

The mass flow rate to the Skultuna area, Sweden, ref. [12] is reduced by 23%. There are other examples where almost similar results were derived when the supply temperature is controlled optimally, taking heat accumulation into account. The supply temperature is also more constant [12].

Fito et al. [25] promote applying an energy-based, demand-oriented algorithm, which recommends rejecting waste heat at 35 °C and combining it with a heat pump and a 40 MWh TES. This design ensures the potential to cover 49% of heat demand with no additional fuel consumption. According to the conclusion presented in [13], a decrease in return temperature by up to 5 °C results in an increase in power and heat production of 400 and 1015 MWh annually, respectively, which are apparently more prominent than we have. Hammer et al. [15] suggest fully charging another artificial energy accumulator to let adjoining supply lines idle and limit the hot water flow rate, which results in a reduced heat loss, and therefore, the simplified calculation shows less energy loss, namely 45.3 kW (heat loss of 82 MWh divided by operation time of 1809 h). On the other hand, both the TES-based and the DH-plant-focused energy-based optimizations do not agree on the same DH system design. Moreover, we suggest changes in the control system only: supplying heat, for example, at 50–60 °C and using the potential of building envelope storage, with no heat pump or TES.

Figure 8 is more about the future application of this research and represents the potential benefits of low-temperature flexible operation if β- and z-values are involved in the supply temperature control.

Figure 8 also indicates that this 80–92 h temp increase was correct and enough not to increase the indoor temperature above 20 °C, although 98–109 h did.

Here, supply temperatures mostly range between 75 and 90 °C (68 °C represents the minimum threshold temperature potentially obtained), whereas reference ones are between 82 and 92 °C. Barone et al. [14] report the same supply temperatures ranging between 90 and 88 °C, where 88 °C represents the minimum threshold temperature obtained at 05:00, while return ones are between 87.5 and 79 °C. Like Saletti et al. [12], we expect the optimal mass flow rate to experience fewer fluctuations during the simulation period, although during the previous management conditions, the operation of the pump changed and varied more sharply.

5. Conclusions

In summary, the supply temperature patterns of six typical buildings in Omsk, Russia, were analyzed using the reference-group-based approach. As DH systems transition to 4GDH, there is significant potential for innovative DH designs that address future distribution network requirements, including efficiency and cost-effectiveness. Building similarities were assessed using pointwise and distributional distance metrics based on indoor temperature data. While the reference-group methodology has proven to be an effective and efficient solution, it can be further enhanced by integrating it into a combined low- and temperature-flexible operation strategy. An analysis based on the stability proportion criterion was conducted to recommend optimal substation valve adjustments and the best material, height, and substation type combinations for office and residential building construction.

By integrating real operational data, mathematical modeling, and optimization techniques, we assess the system’s ability to transition toward LTDH while maintaining efficiency and reliability. The research highlights how supply temperature reductions, when managed effectively, can significantly lower heat losses—up to 20% in some cases—without compromising indoor comfort levels. The findings confirm that flexible temperature operation and heat accumulation strategies can enhance overall system efficiency, reduce operational costs, and prepare the network to integrate renewable energy sources.

One of the key contributions of this research is demonstrating that even in a DH network dominated by older buildings, where high supply temperatures have traditionally been necessary, there is potential for gradual adaptation to lower temperatures. This study provides empirical evidence that, under certain conditions, the supply temperature can be reduced to 50–60 °C without severe disruptions. However, the results highlight challenges, such as improved substation control strategies for mitigating temperature fluctuations and ensuring a smooth transition. Heat accumulation within buildings, particularly in materials with higher thermal inertia, is important for successful LTDH implementation.

Beyond these findings, this study underscores the need for further exploration in several key areas. Future research should focus on refining the control logic of substations, incorporating predictive weather-based adjustments to optimize heat supply dynamically. Additionally, a deeper investigation into the integration of TES at both the building and network levels could further enhance the flexibility and efficiency of DH systems. Another promising avenue is examining how LTDH can be combined with decentralized renewable energy sources and heat pumps, creating a more resilient and sustainable urban heating infrastructure.

Ultimately, this research contributes to the broader discussion on modernizing DH networks, offering valuable insights into the feasibility and challenges of transitioning to LTDH. As cities worldwide seek to improve energy efficiency and reduce carbon footprints, the lessons learned from this case study in Omsk can serve as a foundation for further advancements in DH system optimization and sustainability.