Simplified Energy Model and Multi-Objective Energy Consumption Optimization of a Residential House

Round 1

Reviewer 1 Report

Dear Authors,

The Manuscript presented good quality of research in highlighting, the hypothesis, and methodological statement. The several-past-year period literature sources for the literature analysis are used in the manuscript. The selected problem of the topic „Multi-objective optimization of energy consumption of a residential house“ are interesting. Several comments regarding the improvement of the manuscript are following:

1. Some characterisation with an energy consumption specialisation with a selected type of residential house can be more widely presented. 

2. Some characteristics from the region can by analysed/presented in the manuscript.

3. The equipment for the measured energy consumption characteristics can be presented/mentioned in the manuscript.

4. The requirements of the analysed parameters presented as the documents can be inserted into the literature review part and reference list.

5. The parameters of the measurement of selected parameters (as an example indoor temperature) are expressed in intervals. Are the authors checking possibilities for the calculation using some grey system theory (Deng, 1982) or other methodology for the assessment results values expressed in intervals?

6. For the calculation/optimisation processes the parameters selected with a one optimality direction (minimisation). For the objective optimisation processes the criteria with two optimality directions minimisation and maximisation must be presented.

Reviewer

Author Response

                                                                                                                                                        15^th of September 2022

Dear reviewer,

We appreciate your review, insightful comments and suggestions very much. The paper has been quite a long time in the making and unfortunately, we had to abandon the initial idea of creating a model and addressing the issue of energy demand response. We concentrated on creating a simple model that would be applicable to reduce electricity consumption of a residential house. Therefore, we were also interested in whether we could estimate how the inaccuracy of the weather prediction would affect the predicted electricity price. A similar consideration was how the choice of the desired indoor temperature would affect the price. To avoid the user having to change the desired temperature profiles, we created a system with one parameter, so the user can make a trade-off between thermal comfort and paid price for the electricity. The conversion to the desired temperature is automatic even with respect to the weather forecast.

Below are our reactions to each review point:

1. Some characterisation with an energy consumption specialisation with a selected type of residential house can be more widely presented.

Main electrical appliances are heating (electric), cooling (air conditioning) and charging an electric car. All other appliances have lower power consumption and, like the charging of the electric car they are considered to run in the simulation according to a parameterised schedule. The indoor temperature is next to heating and cooling affected by the weather conditions - outside temperature and solar irradiation. Electricity is taken from the grid and part of it is generated from the solar panels. Redundant electricity can overflow back to the grid. We have limited ourselves to a minimal structure allowing energy balance based on a minimum number of parameters. We want to use the model in the control system of a residential house with restricted memory and computational power. The model will be used for strategic control, but also as a part of expert system allowing the user to make informed decisions. We intend to verify the model's behavior on an ongoing basis and further refine the model based on accumulated experience.

• we have added this in 1. Introduction

2. Some characteristics from the region can by analysed/presented in the manuscript.

The model is primarily designed for a mid-latitude residential house in Europe (where we expect first applications), where the main appliance is heating in winter and cooling in summer. It can be parameterized for other regions. For example, the verification of the sensitivity of energy price predictions to the accuracy of the weather forecasts was performed on data from southern Italy.

• we have added this in 2.1. Mathematical model

3. The equipment for the measured energy consumption characteristics can be presented/mentioned in the manuscript.

In the application, we expect to use standard energy consumption measurements for heating and cooling, electric car, photo voltaic, electrical energy consumption and overflow from and to electric grid. We do not consider measuring heat loss or energy coming from the sun - those energies will be estimated.

4. The requirements of the analysed parameters presented as the documents can be inserted into the literature review part and reference list.

This is the first version of the model with limited information. Once we make the application, we will gradually refine the model to suit the intended use.

5. The parameters of the measurement of selected parameters (as an example indoor temperature) are expressed in intervals. Are the authors checking possibilities for the calculation using some grey system theory (Deng, 1982) or other methodology for the assessment results values expressed in intervals?

Our model is deterministic with stochastic weather forecast as input. Some parameters need not be constants but may depend on other variables or time. Grey systems theory would certainly be an interesting idea to move the model from deterministic behaviour closer to the stochastic nature of the real world.

6. For the calculation/optimisation processes the parameters selected with a one optimality direction (minimisation). For the objective optimisation processes the criteria with two optimality directions minimisation and maximisation must be presented.

In the optimization, we minimized a criterion consisting of two terms - the energy cost and the violation of thermal comfort in the sense of moving away from the desired indoor temperature. In the manuscript, the criterion in equation 14 was incorrectly stated - the term must be in absolute value.  In winter we allowed heating at a lower temperature. Conversely, in summer we cooled to a higher temperature than the user required. Both terms were weighted in the criterion so that the user could choose a trade-off between money paid and comfort.

• we have corrected Equation 14

Sincerely

1. Honc, M. Mrazek, E. Riva Sanseverino, G. Zizzo

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript proposed a simplified mathematical model for estimating the residual energy demand, based on which a multi-objective optimisation is conducted for the trade-off between price and comfort. Conceptually, this is an important issue to deal with. However, the quality of the manuscript must be improved before considered to be acceptable. 

1. The structure of the manuscript must be revised. A comprehensive literature review is missing, with some past works covered in the Introduction. The energy demand forecasting model is not a new topic. The authors are suggested to review similar works and explain the novelty of the model proposed. A separate case study section is also needed to prove the feasibility of the proposed optimisation. 

2. In terms of energy demand optimisation, one problematic premise is that optimisation itself can be seen as a kind of intervention. The model may compromise its accuracy under a scenario that is never been seen before. Therefore, the authors are suggested to discuss this and explain if the parameters (such as thermal loss coefficient) are valid when lower the desired temperature (theoretically yes, but in practice?). 

3. The optimisation solution considers the trade-off between price and comfort at the same time instant. There is also the probability of pre-heating or pre-cooling, which shifts the demand to time slots with lower energy prices. 

Author Response

                                                                                                                                                        15^th of September 2022

Dear reviewer,

We appreciate your review, insightful comments and suggestions very much. The paper has been quite a long time in the making and unfortunately, we had to abandon the initial idea of creating a model and addressing the issue of energy demand response. We concentrated on creating a simple model that would be applicable to reduce electricity consumption of a residential house. Therefore, we were also interested in whether we could estimate how the inaccuracy of the weather prediction would affect the predicted electricity price. A similar consideration was how the choice of the desired indoor temperature would affect the price. To avoid the user having to change the desired temperature profiles, we created a system with one parameter, so the user can make a trade-off between thermal comfort and paid price for the electricity. The conversion to the desired temperature is automatic even with respect to the weather forecast.

Below are our reactions to each review point:

1. The structure of the manuscript must be revised. A comprehensive literature review is missing, with some past works covered in the Introduction. The energy demand forecasting model is not a new topic. The authors are suggested to review similar works and explain the novelty of the model proposed. A separate case study section is also needed to prove the feasibility of the proposed optimisation.

You are right – simple thermal model of a house is not a new thing. The aim is to create a model that predicts the energy consumption of a house based on the weather forecasts, appliances usage and desired indoor temperature. The goal is to create a simple approximation model of the house which can be used in a simple control system with limited memory and computational power. A certain novelty lies in the proposed procedure to allow the user to make a trade-off between thermal comfort and the price paid for electricity using a single parameter. Moreover, this is dependent on the weather forecast. The user or the control system can deviate from the desired indoor temperature and immediately see the predicted effect on the price paid for energy over the forecast horizon.

• we have revised the 1. Introduction of the article, added part about energy models and described key ideas of our solution

As far as the achievability of the proposed optimization is concerned, it is more a question of choosing a desired value of the indoor temperature that may not be achievable in a given situation. In the manuscript, the criterion in equation 14 was incorrectly stated - the distance term from the desired temperature must be in absolute value. We allowed heating at a lower temperature in winter. Conversely, in summer we cooled to a higher temperature than originally required. Both terms were weighted in the criterion so that the user could choose a trade-off between cost and comfort with only one parameter.

• we have added this in 1. Introduction and 5. Conclusion
• we have corrected Equation 14

2. In terms of energy demand optimisation, one problematic premise is that optimisation itself can be seen as a kind of intervention. The model may compromise its accuracy under a scenario that is never been seen before. Therefore, the authors are suggested to discuss this and explain if the parameters (such as thermal loss coefficient) are valid when lower the desired temperature (theoretically yes, but in practice?).

Indeed, our model a very simplified description, the structure and parameters will have to be analysed and fine-tuned during first applications. It may happen that some relationships will not describe the real system accurately enough and will have to be replaced by a more complex description. A certain solution is also the possibility to consider parameters dependent on selected variables or also time-dependent.

• we have added this in 3. Results and discussion and in 5. Conclusion

3. The optimisation solution considers the trade-off between price and comfort at the same time instant. There is also the probability of pre-heating or pre-cooling, which shifts the demand to time slots with lower energy prices.

This could improve the overall energy balance. We have limited ourselves to switching the desired temperature according to the use of the house - night mode, morning and afternoon section and leaving the house section. On the weather forecast horizon, we gave the user the opportunity to make a trade-off between thermal comfort defined by the desired indoor temperature value and the price paid for energy.

• we have added it in 1. Introduction and 5. Conclusion

Sincerely

1. Honc, M. Mrazek, E. Riva Sanseverino, G. Zizzo

Author Response File: Author Response.pdf

Reviewer 3 Report

1. How sensitive are the predictions with respected to the assumptions? Can the authors add a sensitivity analysis as appendix. For clarification to the authors, each system has its own efficiency between 0 and 1. Is determining the accuracy of this efficiency critical to the predicted output of the model?

2. The language requires some modification. Some sentences are too long and can be simplified to improve the understanding

3. To improve understanding of the model suggested in this study, the authors are advised to ass a flow chart just before describing the mathematical model/equations.

4. Could the authors explain why the following parameters were adopted:  The controller’s proportional gain kP was set to 4 and the parameter of the integration time constant Ti was set to 12 000 s. 

Author Response

                                                                                                                                                        15^th of September 2022

Dear reviewer,

We appreciate your review, insightful comments and suggestions very much. The paper has been quite a long time in the making and unfortunately, we had to abandon the initial idea of creating a model and addressing the issue of energy demand response. We concentrated on creating a simple model that would be applicable to reduce electricity consumption of a residential house. Therefore, we were also interested in whether we could estimate how the inaccuracy of the weather prediction would affect the predicted electricity price. A similar consideration was how the choice of the desired indoor temperature would affect the price. To avoid the user having to change the desired temperature profiles, we created a system with one parameter, so the user can make a trade-off between thermal comfort and paid price for the electricity. The conversion to the desired temperature is automatic even with respect to the weather forecast.

Below are our reactions to each review point:

1. How sensitive are the predictions with respected to the assumptions? Can the authors add a sensitivity analysis as appendix. For clarification to the authors, each system has its own efficiency between 0 and 1. Is determining the accuracy of this efficiency critical to the predicted output of the model?

We have chosen model structure as simple as possible to be able to use it in the control system with limited memory and computational power. During the application, we will gradually analyse and refine the model. It is very likely that we will have to replace some of the descriptions with more complex relationships. On the other hand, even the accuracy of the weather forecast is not ideal and will influence predictions and decisions. In this paper we have attempted an sensitivity analysis: how the inaccuracy of the forecast will affect the predicted energy price. It will depend quite a lot on the specific situation and time of the year. A short-term forecast will be much more accurate than longer-term estimates.

• we have mentioned it in 3. Results and discussion and in 5. Conclusion

2. The language requires some modification. Some sentences are too long and can be simplified to improve the understanding.

• we have made a proofreading of our paper

3. To improve understanding of the model suggested in this study, the authors are advised to ass a flow chart just before describing the mathematical model/equations.

• we have added Figure 1 Block diagram of the residential house model structure showing the different subsystems and the links between them into 2.1. Mathematical model

4. Could the authors explain why the following parameters were adopted:  The controller’s proportional gain kP was set to 4 and the parameter of the integration time constant Ti was set to 12 000 s.

The gain of the controller corresponds to the inverse of the system gain, and the integration time constant of the controller is consistent with the dominant time constant of the system. This setting leads approximately to an aperiodic control response. This is made to measure tuning for our hose. We consider either automatic recalculation according to the house parameters or the use of the autotuning methods.

• we have added this to part of the paper where the controller parameters are mentioned

Sincerely

1. Honc, M. Mrazek, E. Riva Sanseverino, G. Zizzo

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Many thanks to the authors for taking into account my comments. The manuscript has been revised accordingly. However, the following comments need to be addressed before the acceptance of this paper. 

1. The numbers of sections are not consecutive. Section 4 is somehow missing. 

2. Instead of separating the methodology and case study sections as most papers do, the authors merged these two parts into section 2. The problem is that readers may feel unconvinced of the outcomes, because there is no holistic evaluation of the proposed simplified model, usually based on a real-life case or simulated case. I wonder why the authors chose not to do this.

3. It is alright not to consider pre-cooling/heating situations in the current study. The control of room temperature is somehow hysteretic, which makes it quite tricky. Do the authors explicitly consider this? For the demand optimisation, the authors are suggested to explain how the optimisation can be implemented in practice. 

4. In the conclusion, the authors summarised three main contributions. From my perspective, the third one (energy demand optimization) looks most relevant. It is suggested the authors can expand this part and use a case study to validate the feasibility of the proposed method in practice. 

Author Response

                                                                                                                                                        23^rd of September 2022

Dear reviewer,

We appreciate your second review very much.

Our reactions to each review point:

1. The numbers of sections are not consecutive. Section 4 is somehow missing.

• you are right, we have corrected this

2. Instead of separating the methodology and case study sections as most papers do, the authors merged these two parts into section 2. The problem is that readers may feel unconvinced of the outcomes, because there is no holistic evaluation of the proposed simplified model, usually based on a real-life case or simulated case. I wonder why the authors chose not to do this.

• it would be better to separate model equations and definitions from simulation results. We are afraid we can’t change this now because the texts are connected and simple separation into separate chapters would make the clarity even worse

3. It is alright not to consider pre-cooling/heating situations in the current study. The control of room temperature is somehow hysteretic, which makes it quite tricky. Do the authors explicitly consider this? For the demand optimisation, the authors are suggested to explain how the optimisation can be implemented in practice.

• we do consider dynamics of the house and dynamics of the temperature controllers. This should resemble real house key behaviour
• implementation in practice is possible through standard numerical optimization methods, we have added this in the conclusion

4. In the conclusion, the authors summarised three main contributions. From my perspective, the third one (energy demand optimization) looks most relevant. It is suggested the authors can expand this part and use a case study to validate the feasibility of the proposed method in practice.

• we are sure that the feasibility of the proposed optimization is guaranteed because of our special criteria formulation
• unfortunately, a case study is not possible within the scope of this article. However this is an excellent suggestion and the perfect topic for continued research

Sincerely

1. Honc, M. Mrazek, E. Riva Sanseverino, G. Zizzo

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

As preliminary research, this manuscript gives fair results to the wider audience about this simplified model as well as its usage in optimisation. But from my perspective, a sound research finding should be validated by a strong case study. This work is interesting and I hope to see the following up studies, evaluating the feasibility of the proposed method.