Energy-Saving Operation Control Method for Hot Water Return Pump in Building Automation System

The energy-saving control method optimizes hot water pump operation through schedule-based and heat loss learning controls, reducing energy waste and ensuring system hygiene by adapting to daily demand patterns and tank temperature monitoring.

KR103046426B1Active Publication Date: 2026-11-11유한회사한국기계설비기술
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
KR1020260053171
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2026-03-24
Publication Date
2026-11-11
Estimated Expiration
2046-03-24

AI Technical Summary

Technical Problem

Conventional hot water return pump control methods in building systems face inefficiencies due to continuous operation during varying demand periods, failure to adapt to daily demand fluctuations, and inability to detect temperature abnormalities in the hot water tank, leading to unnecessary energy consumption and potential hygiene issues.

Method used

An energy-saving operation control method that varies pump operation times based on weekly and daily schedules, learns heat loss characteristics, and monitors tank temperature to optimize pump operation and detect abnormalities, using a control unit linked to a building automation system.

Benefits of technology

Significantly reduces power consumption by 40-70% and prevents low-temperature hot water circulation, while detecting equipment anomalies, with minimal initial costs and applicability to existing systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112026035876731-PAT00001_ABST
    Figure 112026035876731-PAT00001_ABST
Patent Text Reader

Abstract

The present invention is an energy-saving operation control method for a return pump (100) of a hot water supply piping system linked to a building automation control system (BAS). By determining the hierarchical schedule of the daily schedule table (311) and the weekday schedule table (312), the operating time (A), preparation time (B), and non-operating time (C) for today are variably set. By independently learning the heat loss time (t_loss) for each section and schedule type, preemptive restart prediction control is performed based on the predicted stop duration (t_stop = α × t_learn). When the tank temperature (Tt) of the hot water tank (30) is below the operating start temperature (T_on), an abnormal alarm and automatic operation review are performed to prevent low-temperature reverse circulation.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to an energy-saving operation control method for a return pump installed to stably supply hot water to a hot water user in a hot water piping system linked to a building automation control system (BAS). More specifically, the invention relates to an energy-saving operation control method for a return pump of a building automation control system that variably sets the daily operating time, preparation time, and non-operating time according to a weekly schedule and a daily schedule, and repeatedly measures, accumulates, and learns the heat loss time required for the hot water temperature in the piping to drop to a set lower limit after the return pump is stopped for each time interval, and predicts and controls the optimal operating cycle and stop duration differentiated for each time interval, thereby minimizing unnecessary operating time of the return pump to significantly reduce power consumption and providing alarm and operation review functions when the temperature of the hot water tank is abnormal. Background Technology

[0002] Generally, hot water supply systems in hotels, hospitals, and large multi-unit buildings store hot water heated in a boiler or heat exchanger in a hot water tank and supply it to each point of use through pipes. Additionally, to prevent a drop in temperature within the pipes even when not in use, the system utilizes a return pump to circulate the hot water back to the tank. The primary purpose of the return pump is to rapidly supply hot water at a set temperature to the points of use, and it performs the critical function of reducing waiting times for hot water at the supply end and maintaining the temperature within the pipes above a certain level.

[0003] The aforementioned hot water tank stores a certain amount of hot water heated by the boiler to perform a buffering function that responds to temporal fluctuations in hot water demand, and hot water recovered by the return pump also flows into the tank to be reheated and supplied again. Therefore, the temperature of the hot water stored in the tank is a key indicator representing the thermal integrity of the entire hot water supply system.

[0004] However, most conventional return pumps operate using a continuous operation method at rated output during building operating hours. This resulted in the problem of unnecessary power consumption, as the pumps continued to operate even when the hot water temperature in the supply pipes was already sufficiently high. In particular, despite the significant fluctuations in supply hot water demand throughout the day, the pumps operated continuously under the same conditions even during periods of low demand, exhibiting an inefficient mode of operation that accounted for a substantial portion of the building's total energy costs.

[0005] In addition, a temperature comparison-based on / off control method (see Patent Documents 1 and 2) has been proposed that detects the hot water temperature in the hot water pipes and stops the return pump when it reaches a reference hot water temperature. However, since this method operates based on a fixed set temperature, it fails to distinguish between fluctuations in hot water demand and differences in heat loss characteristics by time period within the building's operating hours, resulting in energy waste due to excessive operation even during periods of low demand. Furthermore, a method of starting the return pump based on a user's hot water reservation signal (see Patent Document 2) relies on user operation, which has limitations as it is not suitable for automatic operation in large-scale hot water systems such as buildings. Meanwhile, a hot water system that controls tank boiling operation by learning hot water usage patterns by time period from past hot water history (see Patent Document 3) has also been proposed. However, this relates to learning patterns of hot water usage, and its purpose and control target differ from the technology that predicts and controls the operation / stop of the return pump by learning the time required for heat loss in the hot water pipes. Conversely, time-based on / off control using a simple timer failed to reflect actual heat loss characteristics that vary according to season, ambient temperature, and pipe age, leading to a problem where there was a risk of pipe temperatures dropping below hygiene standards.

[0006] Furthermore, conventional hot water pump control methods had a problem in that they could not reflect differences in hot water demand depending on the day of the week. For example, in the case of office buildings, hot water demand is high on weekdays (Monday to Friday) due to a large number of occupants, but demand drops significantly on weekends (Saturday to Sunday) as the number of occupants decreases considerably. In addition, hot water demand patterns differ from those of weekdays or weekends on specific dates, such as public holidays, building closure days, and special event days. However, conventional control methods failed to reflect these differences in demand by day of the week and day of the week, and applied the same operating schedule every day, resulting in energy waste.

[0007] Furthermore, conventional return pump control methods failed to reflect the temperature status of the hot water stored in the tank in the pump operation control. Consequently, even when the tank temperature dropped abnormally due to boiler failure or heat source interruption, the return pump continued to operate according to conventional settings. This resulted in the circulation of low-temperature hot water into the piping system, causing a counterproductive effect that actually lowered the temperature at the end of the system, or posed a risk of hygiene and safety issues due to the failure to detect equipment malfunctions early.

[0008] Accordingly, there is an urgent need to develop an energy-saving operation control method for a hot water return pump that can variably set time zone classifications for each day through weekly and daily schedules in conjunction with a Building Automation System (BAS), learn the actual heat loss characteristics of the piping for each time zone section to predict and control the optimal pump operation / stop cycle differentiated by time zone, and detect temperature abnormalities in the hot water tank to perform appropriate alarms and operation review functions. Prior art literature

[0009] Republic of Korea Registered Patent No. 10-1515971 (April 22, 2015), "Hot water supply facility" Republic of Korea Registered Patent No. 10-1542886 (August 3, 2015), "Method for controlling hot water supply of a hot water supply facility" Republic of Korea Registered Patent No. 10-2316802 (October 19, 2021), "Hot water supply system"

[0010] ASHRAE Handbook - HVAC Systems and Equipment, Chapter 12: Hydronic Heating and Cooling, 2020 Edition Ministry of Land, Infrastructure and Transport, Annex to the "Rules on Technical Standards for Mechanical Equipment," Design Standards for Water Supply and Hot Water Supply Facilities (Design Regulations for Hot Water Circulation Piping and Hot Water Circulation Pumps) The problem to be solved

[0011] The present invention is devised to solve the problems of the prior art as described above, and has the first objective of providing an energy-saving operation control method that minimizes unnecessary operating time of a hot water pump (100) by variably setting the operating time (A), preparation time (B), and non-operating time (C) for each day through a weekly schedule and a daily schedule in conjunction with a building automatic control device (BAS), and repeatedly measuring, accumulating, and learning the time required for heat loss of the piping for each time interval.

[0012] Additionally, the present invention has a second objective of providing a hot water tank temperature abnormality monitoring function that outputs an abnormality alarm and automatically reviews whether to continue operating the return pump (100) when the tank temperature (Tt) measured by the tank temperature sensor (230) installed in the hot water tank (30) drops below the operating start temperature (T_on) of the current time period.

[0013] Furthermore, the third objective of the present invention is to provide a function that detects abnormal fluctuations in the learned heat loss time required to diagnose signs of piping system abnormalities at an early stage.

[0014] Furthermore, the fourth objective of the present invention is to provide an economical operation control method that can be implemented using only a temperature sensor and software logic of a control unit, thereby minimizing initial installation costs and allowing for easy application to existing buildings. means of solving the problem

[0015] The energy-saving operation control method for a return water pump of a building automatic control device according to the present invention for achieving the above objective comprises the following steps: an energy-saving operation control method in which a control unit (300) linked to a building automatic control device (BAS) automatically controls the operation and stop of a return water pump (100) installed in a building hot water supply piping system by combining schedule-based time zone classification and heat loss learning control.

[0016] In the schedule determination step (S50), the control unit (300) receives current date and day of the week information from the clock unit (320) and determines the type of schedule to be applied today by referring to the daily schedule table (311) and the day of the week schedule table (312) that are stored in advance in the memory unit (310). The daily schedule table (311) includes time zone interval information individually set for specific dates (e.g., public holidays, building closure days, special event days, etc.), and the day of the week schedule table (312) includes time zone interval information set for each day of the week from Monday to Sunday. If the current date corresponds to a specific date registered in the daily schedule table (311), the control unit (300) applies the daily schedule with priority, and if it does not correspond, applies the schedule of the day of the week schedule table (312) corresponding to the current day of the week.

[0017] In the time zone classification step (S100), the control unit (300) determines which time zone segment the current time corresponds to among the operating time zone (A), the preparation time zone (B), or the non-operating time zone (C), based on today's schedule determined in the schedule determination step (S50). The operating time zone (A) is a time zone where there is active demand for hot water within the building, the preparation time zone (B) is a transitional segment before and after the operating time zone (A), and the non-operating time zone (C) is a time zone where there is almost no demand for hot water. The start and end times of each time zone segment may be set differently by day of the week or by day.

[0018] For example, in the case of an office building, the weekday schedule table (312) may be set such that the operating hours (A) for weekdays (Mon-Fri) are 09:00 to 18:00, the preparation hours (B) are 07:00 to 09:00 and 18:00 to 20:00, and the non-operating hours (C) are 20:00 to 07:00; and the weekend (Sat-Sun) operating hours (A) are 10:00 to 14:00, the preparation hours (B) are 09:00 to 10:00 and 14:00 to 16:00, and the non-operating hours (C) are 16:00 to 09:00. Additionally, for specific dates such as public holidays, the non-operating hours (C) may be set for the entire 24 hours in the daily schedule table (311), or only the minimum preparation hours (B) may be set to maximize energy saving effects.

[0019] In the temperature detection step (S200), at every preset sampling period (e.g., 10 to 60 seconds), a hot water temperature sensor (210) installed in the hot water pipe (20) measures the hot water temperature (Tr) in real time, a terminal temperature sensor (220) installed at the end of the hot water pipe measures the terminal temperature (Te) in real time, and a tank temperature sensor (230) installed in the hot water tank (30) measures the tank temperature (Tt) in real time.

[0020] In the step of applying the set temperature by time period (S300), the control unit (300) applies the operation stop temperature (T_off) and the operation start temperature (T_on) corresponding to the determined time period interval. In the operation time period (A), the operation stop temperature (T_off_A) is set relatively high (e.g., 58°C to 60°C), and the operation start temperature (T_on_A) is also set relatively high (e.g., 50°C to 53°C) so that hot water at an appropriate temperature is quickly supplied to the hot water usage location. In the non-operation time period (C), the operation stop temperature (T_off_C) is set relatively low (e.g., 53°C to 55°C), and the operation start temperature (T_on_C) is also set relatively low (e.g., 50°C to 52°C), but is maintained above the hygiene safety lower limit temperature (Tmin, e.g., 50°C). In the preparation time period (B), an intermediate set temperature is applied.

[0021] In the driving determination step (S400), the control unit (300) determines whether the number of times learning data is accumulated in the memory unit (310) of the current time interval is greater than or equal to a preset minimum number of learning times (Nmin, e.g., 7 to 14 times), and if it is less than, it automatically switches to a hysteresis control mode based on actual measurement of the hot water temperature (Tr), and if it is greater than or equal to, it automatically switches to a learning-based prediction control mode.

[0022] In the pump stop step (S500), if the return water temperature (Tr) is greater than or equal to the operating stop temperature (T_off) of the current time zone, the control unit (300) stops the return water pump (100) and records the time information of the stop point, the return water temperature (Tr), the identification information of the corresponding time zone, and the current schedule type (by weekday or by day) in the memory unit (310).

[0023] In the step of measuring the heat loss time required (S600), the control unit (300) measures the elapsed time from the point of stopping the hot water pump (100) until the point where the hot water temperature (Tr) drops below the operating start temperature (T_on) of the current time interval as the heat loss time required (t_loss).

[0024] In the heat loss data accumulation step (S700), the control unit (300) accumulates and stores the heat loss required time (t_loss) in the memory unit (310) according to the corresponding time interval (A, B, or C) and the current schedule type. As a result, the heat loss required time measured during the weekday operating time (A) and the heat loss required time measured during the weekend operating time (A) are accumulated independently, so that the difference in hot water demand patterns by day of the week is naturally reflected in the training data.

[0025] In the heat loss pattern learning step (S800), the control unit (300) calculates and updates the learning heat loss time (t_learn) from the heat loss time data accumulated for each time interval and schedule type using the following exponentially weighted moving average formula.

[0026] t_learn(new) = (1 - λ) × t_learn(existing) + λ × t_loss(latest)

[0027] Here, λ is the learning rate, 0 < λ < 1, and preferably set to 0.1 to 0.3. The learning rate (λ) is a weighting coefficient that determines the ratio in which the most recently measured heat loss time (t_loss) is reflected in the learning heat loss time (t_learn). The larger the value of λ, the higher the reflection ratio of recent data, allowing for rapid adaptation to fluctuations in hot water demand; conversely, the smaller the value of λ, the more the cumulative effect of past data is maintained, resulting in stable learning. For example, when λ is set to 0.2, the most recent single measurement data is reflected in the learning value by 20%, the previous second data is reflected by 16% (= 0.2 × 0.8), and the previous third data is reflected by 12.8% (= 0.2 × 0.8²), so that data within the last five measurements accounts for approximately 67% of the total learning value. This means that under conditions where the hot water pump (100) repeats a stop-start cycle an average of 5 to 10 times per day, the recent heat loss pattern of about 1 to 2 days is primarily reflected in the learning, providing a temporal resolution that can adaptively respond to seasonal changes or gradual changes in building usage patterns. As a result, the time required for learning heat loss during weekday operating hours and the time required for learning heat loss during weekend operating hours are learned independently, respectively, so that it can adaptively respond to differences in hot water demand by day of the week.

[0028] In the predictive operation step (S900), the control unit (300) calculates the predicted stop duration (t_stop = α × t_learn) by applying a safety factor (α, 0 < α < 1) to the learning heat loss time (t_learn) corresponding to the current schedule type and time interval, and preemptively restarts the return pump (100) before the return temperature (Tr) reaches the operation start temperature (T_on) when the predicted stop duration (t_stop) has elapsed.

[0029] In the pre-warming stage (S1000), during the preparation time period (B), in the section immediately before entering the operating time period (A), the control unit (300) performs a pre-warming operation by starting the return pump (100) in advance to raise the hot water temperature in the piping to the level of the operating stop temperature (T_off_A) of the operating time period (A). As a result, hot water at a set temperature can be immediately supplied to the hot water usage point from the start of the operating time period (A).

[0030] In the hot water tank temperature abnormality monitoring step (S1100), the control unit (300) compares the tank temperature (Tt) received from the tank temperature sensor (230) with the operating start temperature (T_on) of the current time interval. If the tank temperature (Tt) drops below the operating start temperature (T_on), the control unit (300) determines this as an abnormality in the heat source supply of the hot water tank (30), immediately outputs a hot water tank abnormality alarm signal to the BAS or field alarm device, stops the operation of the return pump (100), or switches to a tank temperature recovery standby mode, thereby suspending the operation of the return pump (100) until the tank temperature (Tt) recovers above the operating start temperature (T_on). This is intended to prevent the adverse effect of the low-temperature hot water inside the hot water tank (30) circulating to the hot water supply piping system and thereby lowering the hot water temperature within the piping. When the tank temperature (Tt) recovers above the operating start temperature (T_on), the abnormality alarm is released, and the system returns to a normal schedule-based learning control logic.

[0031] In the safety monitoring stage (S1200), the control unit (300) continuously monitors the end temperature (Te), and if the end temperature (Te) drops below the hygiene safety lower limit temperature (Tmin, e.g., 50℃), it performs a safety interlock function that immediately forces the return pump (100) to start and outputs an alarm signal, regardless of the current time zone and operating status. Here, the hygiene safety lower limit temperature (Tmin) is set based on the end temperature (Te), and the measurement location and control purpose are different from the operating start temperature (T_on) which is set based on the return temperature (Tr) of the return pipe (20). However, in the above hot water tank temperature abnormality monitoring step (S1100), when the tank temperature (Tt) is below the operating start temperature (T_on), the forced start of the return pump (100) may cause a reverse effect of low-temperature hot water reverse circulation. Therefore, the control unit (300) simultaneously outputs a hot water tank abnormality alarm and a terminal temperature risk alarm instead of the forced start of the return pump (100) to induce prompt action by the manager. When the terminal temperature (Te) recovers to a temperature above the hygienic safety lower limit temperature (Tmin) by a preset safe return temperature difference (e.g., 5℃), it returns to a normal schedule-based learning control logic. Effects of the invention

[0032] The energy-saving operation control method for a hot water pump of a building automatic control device according to the present invention provides a significant energy-saving effect that can reduce the power consumption of the hot water pump (100) by 40% to 70% compared to conventional continuous operation time by applying a time zone classification optimized for different hot water demand patterns for weekdays / weekends / holidays and specific dates through hierarchical schedule management by a weekday schedule table (312) and a daily schedule table (311), and by independently learning and predicting the time required for heat loss of the piping for each time zone section and schedule type.

[0033] In particular, on weekends or public holidays, the proportion of non-operating hours (C) increases, so the pump stop time is automatically extended, and on days off, the entire 24 hours can be set as non-operating hours (C) according to the daily schedule, thereby further maximizing the energy saving effect.

[0034] In addition, the system prevents reverse circulation of low-temperature hot water caused by boiler failure, etc., through an abnormal alarm and automatic operation review function based on monitoring the tank temperature (Tt) of the hot water tank (30), and enables preventive maintenance through a piping equipment abnormality diagnosis function based on detection of abnormal fluctuations in the learned heat loss time.

[0035] Furthermore, it can be implemented using only the software learning logic of the temperature sensors (210, 220, 230) and the control unit (300), and since it is linked with a building automation system (BAS) to enable schedule setting, remote monitoring, and operation parameter changes, it has the economic advantage of having very low initial installation costs and being applicable to existing buildings with minimal construction. Brief explanation of the drawing

[0036] FIG. 1 is an overall system configuration diagram of a hot water piping system to which the energy-saving operation control method for a hot water pump of a building automatic control device according to the present invention is applied. FIG. 2 is a conceptual diagram showing the hierarchical structure of schedule determination and time zone classification (daily schedule → weekday schedule → time zone interval) according to the present invention. FIG. 3 is a timeline comparison diagram showing an example of the application of a schedule by day of the week according to the present invention (weekdays vs. weekends). FIG. 4 is a flowchart showing the overall control flow of the energy-saving operation control method for a hot water pump according to the present invention. Figure 5 is a timing diagram showing the process of measuring and learning the time required for heat loss by time interval according to the present invention. Figure 6 is a graph comparing the distribution of learning heat loss time by time interval according to the present invention. Figure 7 is a graph comparing the pump operation / stop pattern by time period in the predictive operation mode according to the present invention with the conventional continuous operation method. FIG. 8 is a timing diagram showing the operation of a pre-warming operation in a preparation time period (B) according to the present invention. FIG. 9 is a flowchart showing the logic for monitoring abnormal temperature of a hot water tank and reviewing operation according to the present invention. Specific details for implementing the invention

[0037] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

[0038] Referring to FIG. 1, the hot water supply piping system according to the present invention comprises a hot water tank (30) that stores hot water heated in a boiler or heat exchanger (not shown), a hot water supply pipe (10) that supplies hot water from the hot water tank (30) to each place of use, a return pipe (20) that recovers unused hot water to the hot water tank (30), a return pump (100) installed in the return pipe (20) that circulates hot water to supply hot water to places of use, a return temperature sensor (210), an end temperature sensor (220), a tank temperature sensor (230), and a control unit (300). The above hot water tank (30) stores a certain amount (e.g., 500L to 5,000L) of hot water heated in a boiler to perform a buffering function to respond to temporal fluctuations in hot water demand, and hot water recovered by the return pump (100) also flows into the hot water tank (30) and is supplied again after reheating.

[0039] The above control unit (300) can be implemented as a PLC, DDC, or dedicated temperature controller and includes a memory unit (310) that stores heat loss time data and a schedule table, and a clock unit (320) that provides real-time date, day of the week, and time information. The memory unit (310) is composed of non-volatile memory (e.g., EEPROM or Flash Memory) and stores a daily schedule table (311) and a weekday schedule table (312). The control unit (300) controls rated operation and complete shutdown by outputting an operation / stop contact signal to a magnetic contactor (not shown) of the hot water pump (100). In addition, the control unit (300) is linked with a building automation system (BAS) to provide schedule setting, remote monitoring, and data logging functions.

[0040] Referring to FIG. 2, the schedule management according to the present invention is composed of a three-layer structure of daily schedules (top level), weekday schedules (basic level), and time zone intervals (lower level). The control unit (300) checks the current date and weekday from the clock unit (320) at midnight or before the start of the workday each day, and first checks whether today is registered in the daily schedule table (311). If it is registered, the corresponding daily schedule is applied as the top priority, and if it is not registered, the schedule of the weekday schedule table (312) corresponding to the current weekday is applied. Through this hierarchical determination, the manager can set a regular weekday driving pattern as the basic level, while individually overriding the schedule for holidays or special days.

[0041] Referring to FIG. 3, in an example of a daily schedule for an office building, on weekdays (Mon-Fri), the operating time (A) is set to 09:00 to 18:00 (9 hours), the preparation time (B) to 07:00 to 09:00 and 18:00 to 20:00 (4 hours), and the non-operating time (C) to 20:00 to 07:00 (11 hours). On the other hand, on weekends (Sat-Sun), the operating time (A) is reduced to 10:00 to 14:00 (4 hours), the preparation time (B) to 09:00 to 10:00 and 14:00 to 16:00 (3 hours), and the non-operating time (C) to 16:00 to 09:00 (17 hours). As a result, on weekends, the proportion of the non-operating time (C) increases from 46% on weekdays to 71%, so the stop time of the circulation pump (100) is automatically extended and the energy saving effect is maximized.

[0042] Additionally, if a public holiday (e.g., January 1, Lunar New Year, Chuseok, etc.) or a building closure day is registered in the daily schedule table (311), the schedule for that day may set the non-operating time period (C) to the entire 24 hours, or set only the minimum preparation time period (B) (e.g., 1 hour from 08:00 to 09:00) to perform only the minimum circulation to maintain the pipe temperature. Conversely, if a special event day within the building is registered, the operating time period (A) may be extended beyond the usual (e.g., from 07:00 to 22:00) to provide extended hot water supply.

[0043] Referring to FIG. 4, in the overall control flow, the schedule determination step (S50) is positioned at the top to first determine today's schedule type, and then heat loss learning-based control is performed below the time zone classification step (S100). In the temperature detection step (S200), the control unit (300) removes measurement noise by performing moving average filtering on the return water temperature (Tr), end temperature (Te), and tank temperature (Tt) data.

[0044] Referring to FIG. 5, in the heat loss time measurement step (S600), the time is measured as 30 to 60 minutes in the non-operation time period (C) and 8 to 20 minutes in the operation time period (A). In the heat loss data accumulation step (S700), the most recent N times (e.g., 14 to 30 times) of data is stored in the memory unit (310) in independent ring buffers for each time period interval and schedule type.

[0045] Referring to Fig. 6, in the heat loss pattern learning step (S800), t_learn ≈ 12 minutes for weekday operating hours (A), t_learn ≈ 25 minutes for weekend operating hours (A), and t_learn ≈ 45 minutes for non-operating hours (C) are each learned independently.

[0046] The update process of the learning heat loss time (t_learn) is explained in detail using the non-operational time period (C) as an example as follows. When the learning rate λ is set to 0.2 with the initial learning value t_learn(existing) = 40 minutes, if t_loss = 42 minutes is measured in the first measurement, t_learn(new) is updated to t_learn(new) = (1 - 0.2) × 40 + 0.2 × 42 = 32 + 8.4 = 40.4 minutes. Next, when t_loss = 38 minutes is measured in the second measurement, t_learn(new) is updated to (1 - 0.2) × 40.4 + 0.2 × 38 = 32.32 + 7.6 = 39.92 minutes, and when t_loss = 45 minutes is measured in the third measurement, t_learn(new) is updated to (1 - 0.2) × 39.92 + 0.2 × 45 = 31.94 + 9.0 = 40.94 minutes. In this way, the learning heat loss time (t_learn) gradually converges by reflecting recent data at a rate of λ for each measurement, preventing excessive fluctuations in the learning value caused by sudden outliers.

[0047] According to a preferred embodiment of the present invention, the learning rate (λ) can be set differentially for each time interval. Specifically, during the operating time period (A), since there is frequent outflow of water due to hot water usage and thus large fluctuations in the heat loss pattern, the learning rate (λ_A) is set relatively large (e.g., 0.2 to 0.3) to allow for rapid adaptation to recent heat loss fluctuations. On the other hand, during the non-operating time period (C), since there is almost no hot water usage and heat loss occurs mainly through natural heat dissipation through the outer wall of the pipe, the heat loss pattern is relatively stable. Therefore, the learning rate (λ_C) for the non-operating time period (C) is set relatively small (e.g., 0.1 to 0.15) to maintain the cumulative stability of past data while allowing for gradual adaptation only to gradual fluctuations caused by seasonal changes. The learning rate (λ_B) for the preparation time period (B) is set to an intermediate level between the two (e.g., 0.15 to 0.2). Through these differential learning rates for each time interval, rapid adaptability is secured during periods of large fluctuations in hot water demand, and learning stability is secured during periods of stable hot water demand, thereby achieving optimal prediction accuracy overall.

[0048] In the predictive operation step (S900), a predicted stop duration (t_stop = α × t_learn) is calculated after applying a safety factor (α, preferably 0.7 to 0.9). In the predictive control mode, a real-measure-based safety device (Failsafe) is also used to immediately start the system if the return temperature (Tr) drops below the operating start temperature (T_on).

[0049] Referring to FIG. 8, in the preceding warm-up phase (S1000), in the section immediately before entering the operating time (A) during the preparation time (B), the control unit (300) starts the return water pump (100) when the return water temperature (Tr) is lower than the operating stop temperature (T_off_A) of the operating time (A) to circulate hot water through the hot water supply pipe (10) and the return water pipe (20) to raise the pipe temperature to the level of T_off_A.

[0050] Referring to FIG. 9, in the hot water tank temperature abnormality monitoring step (S1100), the control unit (300) compares the tank temperature (Tt) with the operating start temperature (T_on), and if it is lower than or equal to, outputs an abnormality alarm and stops the return pump (100) to switch to a tank temperature return standby mode. In addition, as a facility abnormality diagnosis function, if the condition in which the heat loss time (t_loss) is shortened to less than or equal to a preset abnormality detection ratio (e.g., 70%) compared to the learning heat loss time (t_learn) persists for three consecutive times or more, it is determined that there is a piping facility abnormality and outputs a diagnosis alarm.

[0051] Referring to Fig. 7, on weekdays, the operating rate is approximately 12% to 22% during non-operating hours (C, 11 hours) and approximately 33% to 50% during operating hours (A, 9 hours), and the 24-hour average operating rate decreases to approximately 25% to 40%. On weekends, the non-operating hours (C) are extended to 17 hours, further reducing the 24-hour average operating rate to approximately 15% to 30%, and on public holidays (24 hours of non-operating), the operating rate decreases to approximately 10% to 20%, making it possible to reduce power consumption by 60% to 85% compared to conventional continuous operation during operating hours.

[0052] The operation scenario of the control system according to the present invention is specifically described in terms of the learning progress stages as follows.

[0053] In the initial learning phase (system initial operation ~ less than Nmin of accumulated data), since there is insufficient heat loss data for the corresponding time interval, the control unit (300) operates in a hysteresis control mode. In this phase, basic on / off control is performed in which the return pump (100) stops when the return temperature (Tr) reaches the operation stop temperature (T_off), and restarts when the return temperature (Tr) drops to the operation start temperature (T_on) due to heat loss in the piping. During this process, the time required for heat loss (t_loss) is measured and accumulated for each stop-start cycle, and learning data is accumulated. For example, if the return pump (100) repeats the stop-start cycle 8 times on average per day, the minimum number of learning cycles (Nmin = 7 times) is satisfied within about 1 day.

[0054] In the learning transition phase (when the number of accumulated cycles reaches N min or more), the control unit (300) automatically switches from a hysteresis control mode to a predictive control mode for the corresponding time interval. At this point, since the learning heat loss time (t_learn) may not yet have sufficiently converged, a margin based on the safety factor (α) is important. For example, if the learning heat loss time (t_learn) is 40 minutes and the safety factor α = 0.8, the predicted stop duration (t_stop = 32 minutes) is set, ensuring a safety margin of 8 minutes relative to the actual heat loss time (40 minutes). In addition, a real-measure-based safety device (Failsafe) is used in parallel to immediately start the system if the return water temperature (Tr) drops below the operating start temperature (T_on) even before the predicted stop duration (t_stop) has elapsed, thereby preventing hygiene and safety issues caused by prediction errors in the early stages of learning.

[0055] In the learning stabilization stage (after the number of accumulated cycles reaches more than twice Nmin), the learning heat loss time (t_learn) has sufficiently converged to the actual heat loss characteristics of the corresponding time interval. In this stage, the accuracy of the predicted stop duration (t_stop) is increased, so the frequency of the actual measurement-based safety device (Failsafe) being triggered is significantly reduced, and the stop time of the circulation pump (100) is optimized, thereby maximizing the energy saving effect. Although the pipe heat loss characteristics gradually change as the season changes, the learning heat loss time (t_learn) automatically follows the exponentially weighted moving average, so continuous operation throughout the four seasons is possible without separate operation by the manager.

[0056] The basis for the quantitative calculation of the above energy saving rate is explained as follows, using the specific conditions of an office building as an example. The building conditions are assumed to be a gross floor area of ​​10,000 m², a total length of hot water piping of 500 m, a pipe diameter of 50 A, an insulation thickness of 25 mm (glass wool), and a rated output of 0.75 kW for the circulation pump.

[0057] In the conventional method (continuous operation during operating hours), when the hot water pump (100) is operated continuously during the building's operating hours (weekdays 09:00 to 18:00, 9 hours), the daily power consumption is 0.75kW × 9 hours = 6.75kWh, and the annual power consumption (based on 250 weekdays) is 6.75 × 250 = 1,687.5kWh.

[0058] When the present invention is applied (learning-based predictive control), the learning heat loss time required (t_learn_A ≈ 12 minutes) in the operating time period (A), and with a safety factor α = 0.8 applied, the predicted stop duration (t_stop ≈ 10 minutes) is 10 minutes, and it takes about 5 minutes to raise the return water temperature (Tr) to T_off when the pump is started once. Therefore, one operating cycle is about 15 minutes (5 minutes of operation + 10 minutes of stop), so about 36 operating cycles are repeated during the 9 hours (540 minutes) of the operating time period (A), the actual operating time is about 180 minutes (3 hours), and the operating rate is about 33%. In the non-operating time period (C), t_learn_C ≈ 45 minutes and t_stop ≈ 36 minutes, so the operating rate decreases to about 12%. The 24-hour weighted average operating rate, including the 4-hour preparation time (B), is calculated to be approximately 28%.

[0059] Accordingly, when the present invention is applied, the daily power consumption on weekdays is 0.75kW × 9 hours × 0.28 = 1.89kWh, and the annual power consumption (250 weekdays) is approximately 472.5kWh. On weekends (104 days per year), the operating rate decreases further to approximately 0.75kW × 4 hours × 0.20 = 0.6kWh / day, and 62.4kWh per year. On public holidays (16 days per year), it is approximately 0.75kW × 1 hour × 0.15 = 0.11kWh / day, and 1.8kWh per year. Therefore, when the present invention is applied, the total annual power consumption is approximately 536.7kWh, and a reduction in power consumption of approximately 68% is achieved compared to the annual power consumption of 1,687.5kWh of the conventional method.

[0060] Meanwhile, the basis for setting the anomaly detection rate to 70% in the above-mentioned piping system anomaly diagnosis function is as follows. In a normal state, the time required for heat loss (t_loss) exhibits a natural variation range of approximately ±20% relative to the time required for learning heat loss (t_learn) due to fluctuations in ambient temperature, hot water usage, etc. That is, under normal operating conditions, the ratio of t_loss / t_learn typically ranges from 0.8 to 1.2. On the other hand, if an equipment anomaly occurs, such as the detachment of pipe insulation, valve leakage, or pipe damage, the pipe heat loss increases abnormally, and t_loss is significantly shortened to 70% or less of t_learn (i.e., t_loss / t_learn ≤ 0.7). Therefore, the anomaly detection rate of 70% is set with a sufficient margin (10%p) above the lower limit (80%) of the normal variation range (80%~120%), and is a threshold value capable of detecting actual equipment anomalies while minimizing false positives. In addition, if an alarm is triggered based on only one abnormality detection, there is a possibility of false positives due to factors such as a temporary increase in hot water usage, so the reliability of the diagnosis is ensured by adding a condition of continuing for three or more consecutive times. Explanation of the symbols

[0061] 10 : Hot water supply pipe 20 : Hot water circulation pipe 30 : Hot water tank 100 : Return pump 210 : Hot water temperature sensor 220 : Terminal temperature sensor 230 : Tank temperature sensor 300 : Control unit 310: Memory section 311 : Daily Schedule Table 312 : Schedule Table by Day of the Week 320 : Clockwork A: Operating hours B: Preparation time C: Non-operating hours S50: Schedule determination step S100: Time zone classification step S200: Temperature sensing step S300: Time-based set temperature application step S400: Driving determination stage S500: Pump stop step S600: Heat loss time measurement step S700: Heat loss data accumulation stage S800: Heat loss pattern learning stage S900: Predictive driving stage S1000: Preliminary warm-up phase S1100: Hot water tank temperature abnormality monitoring stage S1200: Safety monitoring stage Tr: Return bath temperature Te: Terminal temperature Tt: Tank temperature T_off, T_on: Operation stop / start temperature Tmin: Hygiene safety lower limit temperature (based on terminal temperature Te, 50℃) t_loss : Time required for heat loss t_learn : Time required for training heat loss t_stop : Predicted stop duration α : Safety factor λ : Learning rate λ_A, λ_B, λ_C: Differential learning rates by time interval Nmin: Minimum number of training iterations

Claims

Claim 1 In an energy-saving operation control method for a return pump (100) installed to supply hot water to a hot water user in a hot water supply piping system of a building, the control unit (300) receives current date and day of the week information from a clock unit (320), and determines the type of schedule to be applied today by referring to a daily schedule table (311) and a day of the week schedule table (312) stored in a memory unit (310), wherein if the current date is registered in the daily schedule table (311), the daily schedule is applied first, and if it is not registered, the schedule of the day of the week schedule table (312) corresponding to the current day of the week is applied; a time zone classification step (S100) that determines which time zone section the current time corresponds to among the operating time zone (A), preparation time zone (B), or non-operating time zone (C) according to the determined schedule; and a return temperature sensor (210) installed in the return pipe (20) measures the return temperature (Tr). A temperature detection step (S200) in which a terminal temperature sensor (220) measures the terminal temperature (Te) and a tank temperature sensor (230) installed in the hot water tank (30) measures the tank temperature (Tt); a time-specific set temperature application step (S300) in which the control unit (300) applies an operation stop temperature (T_off) and an operation start temperature (T_on) corresponding to the determined time-time interval, wherein in the operation time-time (A), the operation stop temperature (T_off_A) is set relatively high and the operation start temperature (T_on_A) is also set relatively high, in the non-operation time-time (C), the operation stop temperature (T_off_C) is set relatively low and the operation start temperature (T_on_C) is also set relatively low, and in the preparation time-time (B), an intermediate set temperature is applied; and an operation determination step (S400) in which the control unit (300) automatically switches between a hysteresis control mode and a predictive control mode according to the number of times learning data is accumulated in the memory unit (310) of the current time-time interval.A pump stop step (S500) for stopping the recirculation pump (100) and recording stop time information and schedule type when the recirculation temperature (Tr) is greater than or equal to the operating stop temperature (T_off) of the current time interval; a heat loss time measurement step (S600) for measuring the elapsed time from the time the recirculation pump (100) stops until the time the recirculation temperature (Tr) drops below the operating start temperature (T_on) of the current time interval as the heat loss time (t_loss); a heat loss data accumulation step (S700) for accumulating and storing the heat loss time (t_loss) in a memory unit (310) for each time interval and current schedule type; a heat loss pattern learning step (S800) for calculating and updating the learning heat loss time (t_learn) from the heat loss time data accumulated for each time interval and schedule type; and a safety factor (α, 0 < α < A method for controlling the energy-saving operation of a return pump of a building automatic control device, characterized by comprising: a predictive operation step (S900) in which a predicted stop duration (t_stop = α × t_learn) is calculated by applying 1), and the return pump (100) is preemptively restarted before the return temperature (Tr) reaches the operating start temperature (T_on) when the predicted stop duration has elapsed; a hot water tank temperature abnormality monitoring step (S1100) in which, if the tank temperature (Tt) drops below the operating start temperature (T_on), the control unit (300) outputs a hot water tank abnormality alarm and stops the operation of the return pump (100) or switches to a tank temperature return standby mode; and a safety monitoring step (S1200) in which an alarm signal is output when the terminal temperature (Te) drops below the hygiene safety lower limit temperature (Tmin). Claim 2 In claim 1, the daily schedule table (312) is characterized by having the start and end times of the operating time (A), preparation time (B), and non-operating time (C) independently set for each day from Monday to Sunday, the daily schedule table (311) is individually set by overriding the schedule of the daily schedule table (312) for a specific date, and the control unit (300) is linked with the building automatic control device (BAS) to provide a function for remote setting and changing each schedule table by an administrator. Claim 3 A method for controlling the energy-saving operation of a hot water pump in a building automation control device, characterized in that, in the heat loss pattern learning step (S800), the learning heat loss time (t_learn) is calculated independently for each time interval and schedule type by an exponentially weighted moving average formula of t_learn(new) = (1 - λ) × t_learn(existing) + λ × t_loss(latest), wherein λ is a learning rate of 0 < λ < 1. Claim 4 A method for controlling the energy-saving operation of a return pump of a building automatic control device, characterized in that, in the first paragraph, during the section immediately before entering the operating time period (A) in the preparation time period (B), the control unit (300) starts the return pump (100) to raise the hot water temperature in the pipe to the level of T_off_A when the return temperature (Tr) is lower than the operating stop temperature (T_off_A) of the operating time period (A). Claim 5 In claim 1, the method for controlling the energy-saving operation of a hot water pump in a building automatic control device is characterized in that, in the hot water tank temperature abnormality monitoring step (S1100), when the tank temperature (Tt) is below the operating start temperature (T_on) of the current time interval and the terminal temperature danger condition of the safety monitoring step (S1200) occurs simultaneously, the control unit (300) outputs the hot water tank abnormality alarm and the terminal temperature danger alarm simultaneously instead of forcibly starting the hot water pump (100) to induce prompt action by the manager. Claim 6 A method for controlling the energy-saving operation of a circulation pump in a building automatic control device, wherein, in the first paragraph, the control unit (300) further includes a pipe equipment abnormality diagnosis function that outputs a diagnosis alarm signal by determining that the pipe insulation performance has deteriorated or the pipe equipment is abnormal when the newly measured heat loss required time (t_loss) is shortened to less than or equal to a preset abnormality determination ratio compared to the learning heat loss required time (t_learn) of the corresponding time interval for a preset number of consecutive times. Claim 7 A method for controlling the energy-saving operation of a hot water pump in a building automatic control device, wherein, in claim 1, the hot water pump (100) is controlled to operate at a rated rotational speed and completely stop through a magnetic contactor, the memory unit (310) is composed of non-volatile memory so that learning data and schedule tables are preserved even in the event of a power outage, and the control unit (300) is linked with the building automatic control device (BAS) to provide a function for remote monitoring of operation data including learning heat loss time by time interval, pump operating rate, hot water tank temperature history, and operation statistics by schedule type, and a function for remote change of schedule settings.

Citation Information

Patent Citations

  • Noble metal alloy

    JP1995011362A

  • Cogeneration system

    JP2008241145A

  • Combined heat and power supply system

    JP2023141795A

  • Hot water returning system

    KR1020220026108A

  • Heat supply chain management system using artificial intelligence and control method thereof

    KR1020250087930A