Alternatives for natural‐gas‐based heating systems: A quantitative GIS‐based analysis of climate impacts and financial feasibility

The heating of buildings currently produces 6% of global greenhouse gas emissions. Sustainable heating technologies can reduce heating‐related CO2 emissions by up to 90%. We present a Python‐based GIS model to analyze the environmental and financial impact of strategies to reduce heating‐related CO2 emissions of residential buildings. The city‐wide implementation of three alternatives to natural gas are evaluated: high‐temperature heating networks, low‐temperature heating networks, and heat pumps. We find that both lowering the demand for heat and providing more sustainable sources of heat will be necessary to achieve significant CO2‐emission reductions. Of the studied alternatives, only low‐temperature heating networks and heat pumps have the potential to reduce CO2 emissions by 90%. A CO2 tax and an increase in tax on the use of natural gas are potent policy tools to accelerate the adoption of low‐carbon heating technologies.


TA B L E 1 List of abbreviations used in the text
There are three often discussed alternatives to fossil-fuel-based heating systems: high-temperature heating networks (HT, supply temperature of around 85 • C, see Table 1 for a list of abbreviations used in this text), low-temperature heating networks (LT, supply temperature of around 55 • C), and heat pumps (Petrović & Karlsson, 2016;Werner, 2018). Heating networks (also known as district heating) are based on a central heat source and a network of underground water pipes to distribute the heat. Sources of heat include combined heat and power plants (gas, coal, biomass), industrial waste heat and geothermal heat (Lund et al., 2014). Heat pumps use electricity to transfer heat from an outside source, such as the air or water (either stored for this particular purpose or surface water), to a building. There is a wide variety of heat pump technologies available, ranging from small air-to-air heat pumps to large water-to-water heat pumps capable of delivering heat to multiple homes. The Dutch government is considering these three often discussed technologies as viable replacements for the current gas-based heating system (Rijksoverheid, 2017). In this work, we explore their environmental and financial consequences in the context of the Dutch heating transition.
One of the first steps in achieving a reduction in urban heat consumption is improving the insulation of the building stock. Although this will provide a significant reduction of emissions (38-59%), it is not enough to reach the climate goals determined by policy (Buffat et al., 2017;Pinto & Carrilho da Graça, 2018;Werner, 2018).
In literature, we find considerable CO 2 -emission reductions (20-70%) across different alternative heating technologies (Bianco, Scarpa, & Tagliafico, 2017;Delmastro, Mutani, & Schranz, 2016;Lund et al., 2014;Persson & Werner, 2015;Sayegh et al., 2018). However, comparing technologies across different countries is difficult. Their performance is dependent on the climate of the country and the available sources of heat. Furthermore, these comparisons often exclude the effects of increased insulation and the required additional infrastructure. Conclusions about the overall CO 2 reduction potential of alternative heating technologies can only be drawn after a consistent system analysis.
As alternative heating technologies do not operate on natural gas, they require different supporting infrastructures. Heat pumps use the electricity grid, while HT and LT heating networks utilize specific heating networks. This change in infrastructure is generally not taken into account in the assessment of the impact of these technologies. We are aware of only one paper: Love et al. (2017) that established the possible impact of heat pumps on the electricity grid. They found that the peak grid demand could increase substantially as a result of implementing heat pumps.
A large-scale change of the heating system is costly and will require investments over a long period of time. It is therefore critical to assess both the technical and financial feasibility of this transition before making irreversible policy choices. The main factor in establishing the price of heat delivered by heating networks is the cost of infrastructure. Further factors influencing the total cost of a system-wide heating technology replacement are determined by retrofitting buildings and the replacement of the heating technology. Even though financial insight in this transition is crucial for its implementation, system-wide costs including infrastructure are seldom mentioned in the literature (Buffat et al., 2017;Pinto & Carrilho da Graça, 2018;Serrano-Jimenez, Barrios-Padura, & Molina-Huelva, 2017;Werner, 2018).
Research related to alternative heating technologies has focused on individual technological implementations across different countries, making it difficult to compare their CO 2 -emission reduction potentials. This body of literature contributes significantly to the understanding and solving of the multifaceted challenge of the heating transition. However, these models only discuss building refurbishment options and/or a single alternative heating technologies, while the desired situation of the overall heating system could be a combination of these (Bokhoven, 2018).
In addition, the literature has not addressed the system-wide reduction of heating-related CO 2 emissions. The financial implications of sustainable heating technologies are similarly not included in a system-wide analysis. It is therefore unclear how much the emissions related to urban heating can be reduced and what the financial impact of different technologies would be. Especially for home-owners, the heating transition could prove to be an unexpected financial burden (Rijksoverheid, 2017).
We use a bottom-up model based on GIS data to examine the environmental and financial aspects of sustainable heating technologies on a citywide scale. We evaluate HT and LT heating networks with heat pumps as an alternative to natural gas for the heating of residential buildings. Building retrofitting and the different infrastructures required are also included in our analysis. The climate goals set by the Dutch government will be used as a basis for comparison. open source GIS Python package, to analyze multiple GIS datasets. First, two spatial datasets were merged to create one coherent dataset containing building information. Second, the current and future energy consumption of buildings was calculated. Third, this information was used to determine the CO 2 emissions, operating costs, and total investment cost. Finally, the return on investment (ROI) and CO 2 reduction potential in comparison with the existing natural gas network was calculated. An overview of our model is shown in Figure 1, visualizing data flows and sources.

METHODOLOGY
The Dutch government has placed the responsibility for the implementation of this heating transition on its local governments. Current policy plans focus on the replacement of heating technologies on a city-district scale. For the case study of our research, the city of The Hague is used because it represents a typical Dutch city with an old historical center and a variety of building types in its outskirts.

Technologies
The current plan to replace natural gas for the most densely urbanized areas of the Netherlands is mostly based on using large-scale HT heating networks. This network is envisioned to use waste heat from industrial areas to supply heat to multiple cities (known as the warmterotonde, or "heating-roundabout" in the Netherlands). The use of these thermal sources is controversial, as this heat will mostly be sourced from refineries and other fossil-fuel-related industries, potentially creating a technological lock-in with fossil energy sources (Ensoc & RVO, 2018).
The use of water-to-water heat pumps with a 12 kW heating capacity (around 200 m 2 of functional floor area) was assumed. Other heat pump technologies are available with different price ranges. However, these alternatives are more susceptible to extreme cold weather due to their dependency on the outside temperature (Petrović & Karlsson, 2016). Other heating technologies such as pebble heaters, electric resistance heaters, solar boilers, and infrared panels are considered as supporting technology and not capable of fully replacing natural gas as the main heating technology for a building (RVO, 2018).
Our analysis includes the infrastructure transporting the energy from the source to the residential buildings. Infrastructure investment in the electricity grid together with the digging of a well for heat pumps were taken into account. For the heating networks, the implementation of a citywide heating network was assumed. Replacement of the in-house heating system (existing radiators) with an LT heating system was included in the calculation for the LT heating networks and heat pumps. For the HT heating network, we assumed that the old HT heating system remained sufficient as these are often oversized for reliability (Nord, 2016). All assumptions regarding the alternative heating technologies are available in Data S2 in Supporting Information S2.

Calculation of current and future heat demand
Building gas consumption (m 3 /year) was compared with the potential future reduction in this heat demand. Natural gas consumption was converted from m 3 gas to kWh/m 2 on an annual basis. This calculation of the urban energy consumption was based on the paper by Nouvel et al. (2015). Different technologies and their energy sources are simpler to compare using kWh/m 2 . The future heat demand was calculated and based on retrofitted buildings: including an increase in insulation, replacement of the heating technology and heating system.
The future heat consumption of the residential buildings was assumed to be around 70 kWh/m 2 per year after improving the insulation. This is roughly comparable to the average thermal performance the Dutch government aims to achieve for their future built environment (Rijksoverheid, 2017). This improvement in the thermal performance of a building also reduces the impact of a very harsh winter, which some alternative heating technologies are vulnerable to (Werner, 2018). Building heat demand improvements due to insulation were calculated as follows: This difference in heat demand was used to determine the investment cost of insulation. Buildings were not given an increased amount of insulation in the model when consuming less than 70 kWh/m 2 per year.

Calculation of CO 2 reduction potential
In order to determine the CO 2 reduction potential of each alternative heating technology, the change in heating demand through insulation, the replacement of the heating technology, and the efficiency of the corresponding infrastructure was evaluated. Each alternative heating technology was compared to the CO 2 intensity (g CO 2 /kWh) of the existing natural gas system. Based on the reduction potential of each technology, we identified city districts most suitable for a certain alternative heating technology.
A coefficient of performance (COP) was used to describe the energy efficiency of the technology and infrastructure. Heating network transportation losses were assumed to be between 12% and 24%, depending on the technology (Lund et al., 2014). The lower value of 12% was used as the network losses for the LT networks, while the higher value of 24% was used for the HT networks. The heat pumps also have transportation losses from the electricity network, although these will be more marginal (Love et al., 2017). As a result, in the model, the CO 2 reduction is calculated as follows CO 2 reduction where the future heat demand of a building is used in kWh/m 2 /year, COP as the efficiency of the technology and infrastructure, and the CO 2 intensity the CO 2 emissions of the used energy source in comparison with natural gas. The calculation of these CO 2 intensity values is shown in Section 2.2.

Calculation of investment cost and return on investment
For the ROI, the total operating costs of running a natural gas-powered heating system was compared with the required investment and operating cost of the alternative technologies. The total investment cost in € per technology is defined as The investment cost was calculated per building by taking the building retrofitting cost, replacement of in-house heating systems, and the addition of a heat pump and/or heating network infrastructure. An overview of these costs is shown in Data S1 in Supporting Information S2.
Alternative technologies operating costs were based upon replacement costs, consumption of electricity or network heat with the improved insulation, and standing charges. The replacement cost of the boiler, the consumption of natural gas, and the standing charges were included. Current and potential future prices of heat, gas, and electricity were included to predict the influence of changing prices of energy on the overall system.
A ROI per technology was calculated from the payback over 30 years (2020-2050) and the total investment costs.
For the replacement cost, a 15-year lifetime was assumed for both appliances, while for the heating networks a 50-year lifetime was used. For LT and HT heating networks, the infrastructure investments were based on a large-scale heating network project in the Netherlands (CE Delft, 2016). There is however a lack of sources to compare this number with. To illustrate which stakeholder (home-owners and heating network companies/government) will most likely pay for the technology, a breakdown of this investment cost per building was used. In Data S2 in Supporting Information S2, an overview is given of the cost per technology and sources.

Scenarios
Reduction in emissions and the pricing of alternative heating technologies determine their viability as an alternative to natural gas. Development of energy prices and possible governmental interventions influence the affordability and ROI of technologies. Furthermore, the source of heat for each chosen technology influences its CO 2 -emission reduction potential. Its viability as an alternative to natural gas can be explored by looking into potential future developments. For all the technologies, we assume a city-wide implementation.

Available sources of heat
The three mentioned alternative heating technologies operate with different sources of heat. For this analysis, the most widely available sources and potentially sustainable sources of heat available in the Netherlands were evaluated (TNO, 2017). These sources of heat range from grey electricity to the use of PV panels for the heat pumps, geothermal, and sustainable heat sources for the LT heating networks, and CC power generation and HT waste heat for the HT heating networks. An overview of these sources of heat and their CO 2 emissions per kWh of urban heat for heating networks are given in Table 2. These sources of heat were compared based on a direct implementation of the technology and its source of heat.
In the first section of the results these sources of heat are compared in a city-wide implementation for each heating technology based on their current and potential CO 2 reductions. A steady-state implementation from 2020 to2050 was assumed. The average CO 2 production per building in the case study is shown for each technology and source of heat. Additionally, these results are compared with the climate goals for 2030 and 2050 of the Dutch government.

Cost-effectiveness scenarios
Pricing is often used by governments as a method to regulate policy. The alternative heating technologies and possible future interventions of the government should also be included in the analysis to assess their cost-effectiveness. Examples of these pricing methods are: (a) the increase of the price of natural gas to promote the transition to alternative energy technologies, (b) increasing the tax on heat to stimulate the installation of insulation and more energy-efficient heating technologies, and (c) a CO 2 tax to make the reduction of CO 2 emissions more financially attractive. To implement these possible developments in the model, the following scenarios were used: • An increased price of natural gas, 20% and 50% on average until 2050.
• A CO 2 tax of 50 euro per metric ton CO 2 , and 80 euro per metric ton CO 2 (EU, 2016).
• Increased tax on heat with an average increase of 20% and 50% until 2050.
TA B L E 2 CO 2 intensity per kWh of (a) supplied heat for heating networks and heat pumps sources (MRA & TNO, 2017;Stimular, 2016)  An overview of the impact of these scenarios on the input parameters is given in Table 3.
Another aspect of an alternative heating technology is its total investment cost. The build-up of the pricing of each technology is described in Section 2.1.3. It is also possible that the overall cost is higher or lower than we anticipated in this research. To address this uncertainty, we included three investment cost ranges: the standard, low, and high cost. For these low and high ranges, the total alternative heating technologies investment cost is varied with −5,000 and +5,000 euro to account for the uncertainty in technology and infrastructure pricing.
In the second section of the results we compare the ROI across the cost-effectiveness scenarios and the total cost ranges of the alternative heating technologies.

Data selection
The spatial datasets were supplied to us on a ZIP-code 6 level spatial resolution (Dutch ZIP-code). On this spatial resolution, it was not possible to identify the different types of buildings and their age. As a result, the calculation of the future energy demand and potential reduction of CO 2 emissions was aggregated and less accurate for a building-level analysis. Buildings in the dataset with a residential occupancy and existing connection to the gas network were selected. The number of households per ZIP code is derived from the number of existing connections to the gas network. were adjusted with −10% and +10% to determine their impact on the output. The Python code can be found in Supporting Information S1 and the output of the GIS model can be found in Supporting Information S3.

The CO 2 -emission reduction potential
Across all alternative heating technologies, we find that the CO 2 reduction potentials range from 41% to 95% (Figure 2). For HT heating networks, the CO 2 reduction potential ranges from 41% with heat from CC power generation (coal) to 65% with heat coming from waste incineration. For LT heating networks, the residential CO 2 production is reduced by 65% with a geothermal source and 90% when utilizing sustainable waste heat. Heat pumps with "grey" electricity decreases the CO 2 emissions by 60%, which further decreases to 95% when electricity from PV panels is used.
Another important aspect is the CO 2 reduction from the improved insulation of a building. In Figure 2 we show that 33% of the reduction of annual CO 2 is achieved by the improved insulation. This reduction is identical for each technology as they use the same assumptions for the insulation and has nothing to do with the chosen technology.
With the more sustainable sources of heat (lower CO 2 emissions per kWh of supplied heat), LT heating networks and heat pumps are capable of reaching the required 90% reduction in CO 2 emissions. It is also worth mentioning that without the increased insulation, none of the heating technologies and sources will be sufficient for the 2050 climate goal. Besides replacement of the heating technology and source of heat, a reduction in the overall heating demand is required.
Based on the spatial results shown in Figure 3, we identify several city districts particularly suitable for a particular sustainable heating technology. Some districts will still have a relatively high heat demand (mostly older buildings), even after refurbishment, and will therefore be more suited for an HT heating network. The distribution of the CO 2 -emission reduction potential of the LT heating networks and the heat pumps are more evenly matched. The choice for these technologies will have to be based on the availability of local sources of heat. It is most likely that a combination of the technologies will eventually replace the current city-wide natural gas-based system.

Return on investment
The ROI varies from -86% up to 28% across all technologies and scenarios. The ROI, calculated in the model for three different investment ranges and seven future scenarios are shown in Figure 4. We find that in all scenarios the heat pumps have the highest ROI, ranging from −64% in the "high-cost" baseline scenario, up to 28% in the most optimistic "low-cost + 50% price increase for the natural gas" scenario. Also, in this technology, the highest disparity between the different results is found. The LT heating networks ROI ranges from −74% for the high-cost baseline scenario and F I G U R E 3 Annual CO 2 reduction potential for the alternative heating technologies per ZIP code in the city of The Hague (city districts best suited for a certain technology highlighted in orange): (a) LT heating networks; (b) HT heating networks; (c) Heat pumps up to −1% in the low-cost scenario with a 50% price increase for natural gas. The variation of the ROI in the HT heating networks ranges from −86% in the high-cost baseline scenario up to −7% in the low-cost + 50% price of natural gas scenario.
Even with economic incentives, none of the alternative heating technologies has a positive ROI. Only in a low-cost investment range and with a significant increase in the price of natural gas do the heat pumps have the potential to break even or generate a small profit.

Investment costs
Investment costs range from €37,000 to €44,000 between the technologies. Figure 5 provides a comparison of the investment per building for each technology in the standard cost baseline scenario. The investment per building for the heat pumps is the highest with €44,000, but still comparable with the HT heating networks €40,000. LT heating networks require €37,000 per building. From this result and Figure 4 it becomes apparent that although heat pumps have the highest relative payback, only in very specific set of circumstances will this technology have a positive ROI. Implementation across an entire city will require significant investment. In our case study, The Hague, investment costs range from 1.73 billion to 2.91 billion euros. CO₂ tax low CO₂ tax high Price of natural gas +20% Price of natural gas +50% F I G U R E 4 Return on investment over 30 years for each investment range and cost-effectiveness scenario in percentage (higher is better). Underlying data used to create this figure can be found provided in Table S2.3 in Supporting Information S2 The cost attributed to infrastructure improvements differs strongly per technology. For the heat pumps, €10,000 per building is required to improve the electricity network and dig a well. The cost of the heat pumps is the biggest factor in this technology as €23,000 per building is required.
Both the heating network technologies require €26,000 per building to construct the infrastructure. For the heating networks, this is the biggest expense. For each alternative heating technology, the investment in insulation for the case study is €10,700 per building. The technological investment for the LT and HT heating network technologies is between €175 and €4,000 per building. An overview of the investment per technology and subsections can be found in Data S2 in Supporting Information S2.
In the Dutch context, home-owners will pay for insulation and replacement of the heating technology, while the government and energy companies are responsible for infrastructure investments. Therefore, home-owners will be investing €15,000 for the heating network technologies and €34,000 for the heat pump technology. The government and/or energy companies will be investing €26,000 per building for the heating networks scenario and €10,000 for the heat pump scenario. Infrastructure investment has the most influence on the cost of the heating networks, considerably increasing their overall cost.
The results show that even with economic incentives the alternative heating technologies have a difficult business case. Only in the best-case scenario when the heat pumps are cheaper than expected, and with a significant increase in the price of gas will the technology investment generate a small ROI. This means that, in contrast to insulation, the incentive to utilize alternative heating technologies will have to be different for

Sensitivity of the input parameters
Adjusting the input parameters with ±10%, the output of the model varied from +27% and −27%. The price of natural gas is the most influential input parameter with ±27%, while the COP varies the output with ±14% for the ROI. In Figure 6 we show that the price per m 3 of natural gas is the most influential input parameter for this model on the ROI. This corresponds to the results in Section 3.2, where increasing the price of natural gas with +50% leads to the highest ROI. It can also be observed that the COP has a positive influence on the CO 2 reduction. Heat pumps have a high COP in comparison with the other technologies, and consequently the highest CO 2 reduction potential. Additionally, the investment cost of the model is affected by the technology cost (infrastructure, insulations, etc.), and its lifetime.
The results of the sensitivity analysis are in line with the high impact of the price of natural gas on the ROI in Figure 4. The relatively high impact of the COP on the ROI also explains why the heat pumps generate the most ROI of all the alternative heating technologies.

DISCUSSION
Achieving a 90% reduction of CO 2 emissions requires a drastic change in the current Dutch heating infrastructure. This study provides a GIS-based model that clarifies the environmental and financial implications of the Dutch heating transition. We compared the implementation of HT heating networks with LT heating networks and water-to-water heat pumps on a city-wide scale. Besides contributing to the understanding of the 2030 and 2050 climate goals, the financial impact is shown to be of importance for multiple stakeholders in this research.
Through the modeling of the three selected technologies and the evaluation of multiple scenarios, we show that LT heating networks and heat pumps both have the potential to reach the Paris agreement goal (90% reduction of CO 2 emissions before 2050). HT-heating networks could reach the 2030 climate goal (49% reduction of CO 2 emissions), but would significantly limit further reductions. Our findings underline the importance of the sources of heat in reducing the CO 2 impact of residential heating. The adjustment of energy prices or a CO 2 tax has the highest impact on the ROI of the heat pumps and the LT heating networks. With the right policies and tax instruments, they could surpass the break-even point. The results further limit the affordability of the HT heating networks considering they currently even lack taxation. In our model, we also show that the price of natural gas has the highest impact on the ROI of alternative heating technologies. Currently, the energy bill of a Dutch household is largely determined by the consumption of natural gas instead of fixed tariffs.
Increasing the price of natural gas improves the business case for alternative heating technologies significantly, but also makes the cost of urban heating more expensive.
We believe that this research gives some insight into the CO 2 reduction potentials for the Dutch residential building stock. Replacing heating technologies is not sufficient on its own. Acquiring more sustainable sources of urban heat is also required to achieve significant CO 2 reductions before 2050. The development of long-term spatial planning and financial incentives, in cooperation with home-owners, is essential to accelerate this heating transition. Usage of HT industrial waste heat for the 2030 climate goal could limit further reductions in CO 2 emissions and obviate the 2050 climate goal.
In comparison with previous literature, we compared the environmental and financial impact of multiple heating technologies within the same case study. This alleviates the problem of comparing heating technologies across different climates and building types. Also, the inclusion of infrastructure and multiple sources of heat in our analysis gives a broader perspective on the consequences of this adjustment to a heating system.
Although we use GIS data, our results are currently not spatially explicit, beyond visually identifying spatial patterns on a district scale. Further research could identify the buildings most suitable for adjustment to a specific alternative heating technology. For example, a spatially explicit analysis could identify buildings which would be most suitable for HT heating networks. Also, comparing these heating options with further spatial characteristics such as available sources of heat and socio-demographic characteristics would provide a more in-depth spatial analysis. A further development of indicators, and the inclusion of more alternative heating technologies would also improve the outcomes of our model.
A limitation of this study is that we relied on implied data due to a lack of information on heating networks. Especially the infrastructure prices of the heating networks are generally unspecified. The price ranges of the heat pump technology and the chosen technology could also be debated.
We were also unable to include inflation in the model. Last, the embodied energy of alternative heating technologies and their material impact is not included. With more fitting data this methodology can be easily updated and applied to other future scenarios.

CONCLUSION
This study highlights the differences between three main natural gas-free heating technologies on their environmental, technical, and financial aspects. Our main results show that the business cases for the alternative heating technologies is only profitable with the right combination of economic incentives. Without significant subsidies for existing buildings and home-owners, the financial implications could prove fatal for the heating transition.
We show that low temperature (LT) heating networks and heat pumps both have the potential to reduce the Dutch urban heat-related CO 2 output by 90%. A combination of these technologies could be used as an environmental and financial alternative to natural gas. However, these replacement technologies will require a considerable capacity of sustainable sources of heat to reduce CO 2 emissions by 90%.
A combination of policies together with subsidies will give home-owners a strong incentive to refurbish their buildings and lower the residential consumption of heat. In the larger context, our study shows that using industrial HT waste heat for residential urban heating in the warmterotonde will not be sufficient to achieve the 2050 climate goal.
Further research in this direction is encouraged to provide multiple energy evaluation tools for the heating transition. The development of this heating transition could also be influenced by energy storage solutions. At present, the use of energy storage is limited, but in the future, this could have a strong influence on the system (Petrović & Karlsson, 2016). Phase change materials, improvements in battery technology, and localized hydrogen storage present a potentially disruptive development for the overall energy grid (heat and electricity). The material demand for such a large-scale transition of an energy system could also influence or even disrupt critical material supply chains (Sprecher et al., 2017). These developments should be considered in future research on this topic.