Importance of buildings and their influence in control system: a simulation case study with different building standards from Germany

As already mentioned in a lot of instances, the [1] report estimates that around 40% of the world’s final energy is used for residential buildings. The building sector plays a large role in the energy consumption which includes space heating or cooling, domestic hot water and electricity. Buildings with their long lifespan and huge amount of already existing buildings, makes revision in energy characteristics of a building constrained. In many countries, new buildings could be made 70% more efficient than existing buildings [2]. According to [3], the annual rate of new building construction is always higher than the demolition rate. In contrary to the energy policies’ aims of reducing the energy consumption, the energy demand in residential sector has increased by 15% in the last decade in UK [4]. Global energy consumption in the residential sector grows on average at 1% per year [2]. Energy use in the buildings sector follows a trend which increases it by 88% between 2003 and 2050, or 1.4% per year [2]. A whopping 50% of residential sector demand is used for space heating in IEA countries and this is expected to be around 70% for countries such as Germany and Poland [1]. The second major energy demand for residential sector is the electricity. Electricity consumption increases at 2.5% per year, raising its share in the buildings sector from 24% in 2003 to estimated 40% in 2050 [2, p. 200]. Half of all energy saving potential in the residential sector could be from space heating [2]. Water heating also accounts for about 16% of total energy savings in the future [2]. Accordingly, the most effective energy conservation measures are the insulation of external walls (33–60% energy savings), weather proofing of openings (16–21%), the installation of double-glazed windows (14–20%), the regular maintenance of central heating boilers (10–12%), and the installation of solar collectors for sanitary hot water production (50–80%) [5].

The buildings in Germany share a large part of the total energy consumption and since 1979 there has been standards to be followed for new buildings and renovations. According to [3], there are 18.6 million residential buildings in Germany and of which 83% are single or 2 family houses (47% of the German families live in SFH and 2FH). 64% of these single or 2 family houses are built before 1979 and therefore in total there are 65% of residential buildings which are erected before any standard was in place and 39% of those are built between 1950s and 1970s [3]. It is also calculated that only 33% of these old SFHs are renovated [3, 6]. The windows (60%) and the roof (65%) are the most reconstructed parts of the building and still the walls are not given that much of importance with only 35% of the renovations considered to improve the U value of the walls [3]. The basement insulation is the least renovated with 20% [3]. According to [3], there is only 1–1.2% increase in the renovation trend each year and only 4% of these renovations are full renovation of the buildings. SFHs as an average require 152 kWh/m²a end energy and it has just reduced 13% (21 kWh/m²a) in the last 6 years [3]. Dena [3] also showed that 80% of the energy demand in a residential building sector is used in heating and only 15% of this heating demand is delivered by renewable energy while still 46% is via gas. When the installed thermal energy producers are closely compared between 1998 and 2013, the 13% of the oil’s share is now substituted by heat pump and biomass but the gas units are still more or less in the same share of 70% [3]. Surprisingly, [3] also states that in the newly built buildings still 51% are gas boilers and 30% heat pumps and only 1% is solar thermal and the trend is reducing since 2011 for the yearly installed solar thermal systems. The application of renewable systems into buildings is the next step to sustainable future and there are already many research going on in integration of renewable and their benefits in buildings, such are the articles from Enteria et al. [7], which showed how the renewable systems can be integrated into single family houses in Philippines and why all buildings are not the same. Enteria et al. [8] also showed the benefits of utilizing the renewables in single family houses.

The control system for buildings plays an important role and with developments in control strategy around 10–20% of the energy use can be reduced [1]. Regardless of white, gray or black box model of the controllers, the building plays an important role. The crucial part of building modelling in the predictive control is explained in [9]. The building as storage and the energy demand variance between different air tightness and insulation is showed in [10]. Yu et al. [11] listed building system control in three categories: classical control (mainly on/off and PID); soft or intelligent control (based on historical data); and hard or advanced control (based on building model). As described in [12], the model free control strategies use the weather predictions without any building or historical data which manipulate the heating or cooling system or the temperature set points. These are easy to implement, low cost but limited results. The intelligent controllers use a kind of fuzzy logic or a set of rules which are created from the monitoring or current building data [13]. The ability is high, but experienced user is required to set the logical rules and is different for each scenario and the logic is required to be developed from scratch for each building application. The advanced control uses the building model in the control operation. In [12], advanced control is further divided into adaptive, optimal and model predictive control. Adaptive is based on the adjustment of model parameters [14] and optimal controllers find an optimal setting for the parameters which is being adjusted to an optimum of a control strategy [15]. Model predictive control (MPC) does the optimal control with some predictions. Other than these control methods are the non-optimal advanced controls which make the best of the situation even though it is not optimal (e.g., heating the building using electricity when PV is working with certain boundary conditions) [12, 16].

The main parameters usually for a MPC are thermal comfort, peak energy consumption, total energy consumption and applicability of renewable energy sources [16]. The most of the MPCs which are researched for now are specific to that building and may not be the optimal for similar building or application. Due to this the cost of implementing such MPCs are high and the uncertainties further increases when they use the gray or black box models. And for white box models, the building is modelled in a software (TRNSYS, EnergyPlus) in which the optimization has to be carried out in another software (Matlab, GenOpt) and the computing time is high. Hence, in such cases where simulated building model is used for control, they are very specific and the manipulated variables are mostly the set point or on/off of the energy systems. In other cases, the renewable production forecast, the energy system and storage is controlled for optimal operation and the demand forecast is carried out. But in most of the researched cases or implemented MPCs the solution is not universal, thus mass optimization logic is not yet achieved. The necessity of such systems is being showcased in this paper where the control unit can use a learning process to utilize the MPC optimally and manipulate the required variables to achieve an optimal strategy in any kind of building after understanding them and the algorithm will be universally viable which of course can be mass produced and will reduce the implementation cost. But the problem with such universal MPC stay in the demand side where the building’s heating demand is not only different but also fall on the dependency on buildings response, thermal comfort zone, different sizing and components of the energy system. This paper tries to showcase the buildings’ and their standard’s importance in sizing of the energy system. In addition, the control strategy of different building standard differs in accordance to their energy demand, heat storage capability and energy system size.

The role of controllers in building heating or in renewable systems has already been identified. For example, a MPC controller was implemented in a Swiss office building in [17] but they customized the MPC for that particular building and also showcased this causes the cost for such controllers to be too high to be put in use. In addition, MPC was introduced in energy storage system using weather forecast and this was reviewed in the literatures of Thieblemont et al. [12] and Zhun et al. [11]. But what still remains open is the combination of energy systems, the storage and building all together into a single system controller and universalization of this controller. In most of the literature, the MPC is customized and designed especially for the special case and the ultimate aim of the project is to develop a controller hierarchy or a management where all controllers will communicate with each other and work optimally using MPC via energy and demand forecast and learn from the system such that it can be universally adaptive.

The work is partially a part of the project intelligent home energy management (iHEM), more info at [18]. The main objective of this project is to integrate thermal and electrical components with more energy sources at high efficient operation in a household for high self-utilization. For an intelligent control of such system, the load and production energy are forecasted and an optimum predictive operation strategy is found out so that the system is not only efficiently optimized and self-utilization prioritized with maximum possible renewable energy ratio but also economically and strategically viable. The subjective goal of the project is to have an integrated home energy system communicating to all the components through a superior management component. In accordance with this project, in paper [19], the simulation of the novel simulation model of the storage tank with lances is discussed and in [20] the simulation of the entire energy system (building, thermal and electrical system) modelling in TRNSYS is explained along with the assumptions simplifications and limitations for the simulation model. The electrical model has then been developed to a three-phase model. A small description of the simulation is given in the section below but a detailed one can be found in [20]. The previous works done [19‐21] in this project can also provide a background of this work.

The challenge to be addressed here is to develop a universal control logic which could be installed in a house where regardless of renewable system and the building characteristic, the control unit should function optimally and work harmonically in coherence with the energy system and building. The ultimate aim of the project is to build a MPC simulation model with energy system and building in TRNSYS and to develop a control logic using hardware in the loop (HiL) tests. These HiL tests would be used to answer questions like which system configuration works better with which building type or standard with special focus on German buildings and energy demand types.

This paper strives to point out the working inabilities of the current control logics in the current building energy systems and the required advances in that field by pinpointing the differences in the building responses and why all buildings are not the same; what is the required control changes to integrate renewables so as to utilize the renewables effectively and why the control approach for different buildings must vary. In this simulation case study, four different building standards in Germany from vast building standards are taken. These four building types are Sonnenhaus, KfW55, Passive house and WSchVO95. All these house types have different energy demands, different building characteristics and range not only from old house (WSchVO95) to new (KfW55) but also from less energy requirement (Passive house) to high solar energy fraction (Sonnenhaus). The buildings have the same building geometry or structure and only the building envelope’s details are changed according to the requirement of each of the building type. All the buildings are single family houses with same geometry but in accordance with each standard.

The building characteristics of the base case are taken directly from a real KfW55 Building [22]. The KfW55 building is made available from Dammann-Haus GmbH Company and the total U value of the building is 0.18. The other three building models are developed with the same building envelope dimensions and just by changing the characteristics of the walls and the required building standard conditions. As previously mentioned, the paper will focus only on single family house. The house being modelled is with a useable area of 183 m² and a gross volume of 567 m³ with the envelope as shown in Fig. 1. The house has in total two floors and an unheated roof store room. The roof is 63 m² in each side facing North–South and the unheated roof together is 35 m². The floor of this unheated zone is 52 m². The two floors of the heated building are taken as a single zone. The building is structured in SketchUp and is coupled in TRNSYS with the energy system. The fuel cell CHP is modelled in Matlab and is coupled in TRNSYS. The thermal and electrical system with space heating, hot water demand and electricity demand is simulated in TRNSYS.

The building is assumed to be occupied by four people during 22:00–08:00 and by one person during 08:00–13:00 and from 13:00 proportionally the three persons coming in. The shade for the buildings is put on when temperature outside is above 24 °C to reduce solar gains. The windows are also opened for a forced ventilation when temperature is above 26 °C. The building has 9, 7.42, 16 and 1.6 m² window area in south, east, west and north. The more on dimensions and characteristics of the building can be found in [22].

The single family house is assumed to be occupied by four members in all scenarios and the electrical load profile is taken from VDI4655 Norm profile [23]. This profile is being taken from real measurements in one minute intervals. The measurements are done on ten different typical days combining clear and cloudy; summer, winter and transition season; working day and holiday. With those typical days, the annual demand is compiled. The consumption was measured in 15 different climate zones in Germany. The profile is made available in nominated value which has to be customized with the number of occupants and the climate zone. The DHW profile and the electricity demand profile is being developed from this profile. For the simulation the SFH building with 4 occupants is assumed to have 500 kWh per person per year for hot water consumption and a total of 4140 kWh as electrical consumption. More about the demand curve can be found at [23]. For the other standards, the building data are also taken from the real building of that particular standard. Using the KfW55 building geometry, the building simulation model is created. Thus, being easily comparable with one another.

The climate in Ulm is a mid-European cold of a kind. As shown in Fig. 2, a weather data from year 2013 is used for simulation. The temperatures go up to − 15 °C in February winter peak and staying sub-zero for most of the winter as an average during the months November–February. During these months, due to low sun angle and less sun hours not much of solar gains are achieved. But in summer, the average goes above 15 °C with peak mid-day up to 30 °C on some days. The house temperature is set at 20 °C for which the heating control works on with a ± 1 °C and a night temperature reduction of 16 °C during 22:00–06:00.

One more noteworthy thing to reiterate is that the sizing of the energy system is optimized for the base case (KfW55) and the same sizing configuration is used for other three buildings. The main criteria for sizing were high self-production but at the same time economically sensible sizing. The reason for this is to show the effect of sizing and the control system dependency on the building standard. The building KfW55 is selected as a base case because this is one of the currently most built-up single family buildings standard in Germany. The building input parameters for all the buildings along with the main TRNSYS types are mentioned in the table below. The solar thermal is first sized at 26 m² gross collector area with 2 m³ storage tank. While 2 m³ tank was sufficient for KfW55, Sonnenhaus and Passive house but for WSchVO95 a 3.5 m³ storage is required. Along with mere 5 kW gas boiler and 4 kW heating element, the thermal loads of all buildings except WSchVO95 is easily satisfied. The gas boiler makes sure that at 0.75 relative height of the tank (H′) the temperature is 55 °C and flow rate is set at 200 kg/h with the minimum turn down ratio 0.1. For WSchVO95, a 30-kW gas boiler is required and the set point has also to be increased to 70 °C and the mass flow into the gas boiler is also in accordance changed to 1200 kg/h and the sensor for gas boiler control in the storage should be placed at 0.6 H′ only with which the auxiliary volume for the building heating is achieved. In addition, the available 2 kW_el SOFC fuel cell feeds in a maximum of 600 W of thermal gains when temperature of inlet water is less than 30 °C after which the thermal gains gradually decrease and from 60 °C inlet water there are completely no thermal gains from the FC-CHP. As already mentioned, fuel cell is in operation all-round the year and takes the coldest water available in the storage tank and at 90 kg/h cools the fuel cell exhaust for thermal gain. The FC-CHP held in the same conditions for all the building standards.

On the load side, the set point temperature for the hot water is 55 °C and takes the hottest water possible from the tank and mixes it via a diverter–mixer combo and feeds into the fresh water station with a 1500 W/K heat transfer rate to the fresh cold water. For the space heating, there is an active floor area as floor heating in the building model for KfW55 which is 50 m² together for both the heated zone floors. These sum up to a required minimum of 500 kg/h and set maximum of 1000 kg/h for the floor heating mass flow. In addition, in accordance to the ambient temperature, the inlet temperature of the space heating load varies from 25 to 38 °C. As already mentioned, a night temperature reduction is set to 16 °C desired room temperature which is otherwise 20 °C. For passive house and Sonnenhaus also, a floor heating is used. But for those a 100 m² heating area with lesser flow rate of 200–700 kg/h is required. For WSchVO95, the heating temperatures from floor heating are not enough and hence a radiator of 45 kW maximum power with a maximum flow rate of 1200 kg/h with a higher inlet temperature set of 35–57 °C is input.

As in KfW55, there is a heat recovery ventilation system which has exchange effectiveness (E) of 75% at an exchange rate of 0.43 changes per hour. The ventilation and infiltration of the buildings are input according to the standards as shown in Table 2. Moreover, it is assumed to be four people living in the house, and all four spending the night, of which three people go off in the morning and slowly back in the afternoon one by one. 81% of the electrical consumption in the house is assumed to be the other thermal gains in the house. For forced ventilation and shading, the ambient temperature-dependent control opens the jalousie for shading when temperature is above 24 °C outside so that solar gains are avoided inside the buildings and windows are assumed to be opened for forced ventilation when temperature is above 26 °C.

On the electrical side, the system is same in all the buildings except that the excess electrical energy fed to the thermal store is dependent on the temperature level in the store. PV is sized for 2.9 kW_p which occupies 19 m² roof area and the parameters are input according to the data sheet of Aleo Solar S295. The suitable inverter at 3.3 kW, the Stecagrid 3203 from Steca is input. The given inverter efficiency with respect to the power ratio from total inverter size, as shown in the datasheet, is given as a polynomial equation into the inverter model. Battery with 6.8 kWh from BMZ (ESS 7.0) including Steca charge controller with all their properties is input from the datasheets of the respective components. The simulation time step and data files interval are all input as 15-min time step. The matlab fuel cell model is called every time step. The weather data in an hourly dataset are interpolated for 15 min data points.

Even though the sizing of the system is kept same in this case study, the parameters or the control logic have to be fitted with each of the building type. Especially, the auxiliary production units have to be designed according to the space heating demand. Thus, the importance of buildings prove worthy in an energy system design. When the system size also has to be optimized, the controller has to be accordingly improved and this becomes tricky with every other household. Therefore, comes the necessity to use simulations in the design phase and also energy certifications are made as a requirement for each household in most of the countries. But these are done only to know the building demands. It is more important to optimize the space heating demands according to the necessity and utilization intelligently.

When the building responses are closely looked into, it could be seen that the control logic is not sufficient enough. Above in Fig. 10, a winter week’s heating demands of the house is analyzed in detail. On the first graph, the temperature and the global horizontal irradiation are plotted. The temperatures go up to −15 °C on this week and the solar gains are low. And on such a day as expected in the KfW55 house the heating is on almost all the time during the day. By 12:00 pm when there was sun with just up to 400 W/m² during the day, the house got some heat through which was sufficient for the SH to be switched off. But on next day where there was almost no irradiation, the heating had to be switched on with 2 kW power. In the Sonnenhaus, the effect of the solar radiation is not that much while the heat lost through walls are higher and hence the heating system is not uniform. There are also some hystereses on 10 Feb. The building seems to be heated up at the beginning and then compensated by reducing the control back after which again increased. An almost perfect heating can be seen in the passive house where the heating is maximum switched on only for maximum six hour a day in winter. But due to the control logic, it should be noted that the building heats-up up to ca. 5 kW power which for this building is not required. And exact opposite to all these building responses is the WSchVO95, where there are a lot of hystereses and the heating power demand rises up to 10 kW including immediate switch off and on control commands.

Rather than the heating in Winter, the optimisation in transition seasons—spring and autumn (as shown in Fig. 11) are the difficult days to capitalise on. On these days, there are enough renewable energies, cold days but very good opportunities which are at the same time difficult to control. On such days for buildings like WSchVO95, the easy on–off control logic with still sophisticated PID algorithms currently in use be of no good. With 7–8 kW large amount of high grade heat being just circulated in the heating radiator and being returned to the tank creates hysteresis on a daily basis. And talking about the building response, the passive house needs no heating at all on such days. In addition, the Sonnenhaus and KfW55 houses accommodate the changes to an extend and use less heating energy.

In this study, it is shown how large is the difference between buildings of different standards. From the inferences obtained in this simulation, it can be seen that the building plays an important role in the design of the thermal energy system. More than the energy parameters, the reaction and approach to each building type vary. The KfW55 and passive house are quite reactive to the changes in the heating implementation. These both buildings need less time to reach the desired set point temperature, especially in a winter early morning where the set point temperature is varied from 16 to 20 °C after the night reduction for energy savings. But in the similar situation for an old house such as WSChVO95, the heating has to be switched on quite earlier in the morning such as 4 am, if at 7 am the building has to be at 19 °C at least. Moreover, the inlet temperature and the mass flow have to be varied such that the desired room temperature is reached. The approach for a WSchVO95 building is that the inlet temperature has to vary in an increased slope value to reach the required room temperature along with an increased maximum mass flow so that the heat transfer rate is better. But for a passive house or KfW55, the slope of the inlet temperature algorithm has to smaller and the mass flow should be set that not lot of heat is transferred in one go which otherwise will create a hysteresis where the heating unit has to be switched on and off a lot of time. Not only the U value (losses) of the building is to be taken into account, but also the g value (the gains: solar and other) should be considered. For a passive house on a spring day, 3 h of sun in the mid-day will be more than sufficient to keep up the heat and reduce the space heating consumption for the remaining of the day. This can be seen by the distribution of the space heating demand difference between the months for each building type. In good insulated houses, it is evident that the space heating demand is 90% in the 4 winter months. And in the other months, the solar thermal fraction is already above 70%, e.g., passive house.

Without the fuel cell and the excess energy being fed to the thermal storage, a 7–10% increase in solar thermal production is experienced in all four buildings. The constant flow instead of the matched flow would create no change in the gained solar collector energy but with constant flow the achieved temperature is not always optimum which is disadvantage in the summer as there will be less usable high grade water in the storage. The pipes are also losing energy during transport and this is overlooked here and according to bad piping this could already be 10–20% of the total energy demand. In addition, the equal division of the 3-phase loads is not in reality possible but the alternative in real life is having an efficient 3-phase inverter which balances each phase intelligently. Even with such boundary conditions and assumptions, the control logic of the system changes in accordance to the building type. One of the major issues is that all houses in the same building type, which has a different envelope, different U value and a different window size might also not be able to operate efficiently with the same control logic. Moreover, there are a lot of variables in the normal operation of the building which can be simulated but in real life is difficult to control without intelligent controllers. The prerequisite of the intelligent controller can be hugely divided into two: an initial parameter identification which will get info of the building as soon as it is installed in the house via small tests, sensors and user input; then a moving variable which will be gathered from the usage, monitoring system and user validation which will be continuously improved. The first step of an intelligent controller will be able to learn the environment such as building location, building orientation, infiltration, U value, energy losses, g values, solar gains, heating response of the building, energy system components and their size. The continuous improvement of the conditions has to be done to the person gains, ventilation, heating power and slope, grid and gas price, heating response with respect to ambient temperature and energy loss, part of the building with respect to time, etc. With the help of MPC and the parameters read from the system, the control logic should be flexible to take decisions with the forecasted energy demand, and with respect to all these, the energy production should be planned accordingly. As it is seen in the above case study, the transition season is the most important to gain an increased renewable fraction and efficient energy utilization.

As for the differences in the building itself, it can be seen that their responses for the change in the climatic conditions are different. For the space heating, the passive house is the best, but when closely looked into, it can be noticed for such a house the SH power cannot be as high as a WSchVO95 while the building cannot store a lot of heat due to the heavy insulation and airtightness. In addition, the ventilation is more important in passive house and KfW55 than other two standards while it is forced ventilation and should take care of the comfort of the persons living there. Sonnenhaus is also a good efficient house but for some small discrepancies, these houses are not as efficient as KfW55 and passive house which is already discussed in section ‘Sonnenhaus results’. In terms of control logic, a lot of changes have to be made between the different buildings. For example, as seen here, the gas boiler sensor in the tank, the mass flow and the set point temperature had to be changed according to building type. These are due to the account of the changes in the mass flow in the floor/radiator heating along with the power required by the space heating. The comparison of the building and their optimized energy system shows that for each of the building type, the response and demand of the building is different. Thus, the control strategy is important and a simple on/off controller does not satisfy the needs. With this simulation study, the importance of buildings in an energy system is emphasized.

This simulation case study pronounces the importance of buildings in the energy system design and why space heatings are a node point for a better efficient system. This is difficult since buildings are not always the same and react differently for different scenarios and building types, thus the necessity of an intelligent control system for this purpose is high. To move over to the next step of efficiency, the transition seasons energy production and demand management needs to be more precise, this can be done cleverly with self-learning controllers with continuous improvement which uses the service of monitoring, MPC, energy and demand forecast and user inputs. In addition, the MPC control unit should be universal so that it can adapt to the system locally.

As in the paper, a demonstrator system with all the components is built up at Sailer GmbH’s premises. The system will be emulated for a single family house demand with a kind of HiL setup where the control logic of the system will be developed and tested. Thus, the different approaches for different buildings where a single flexible control logic will adapt to the system can be developed. To further develop the model, the simulation will be compare with the real-time data from the demonstrator and the simulation model will be tuned for a much precise operation.