Orig­i­nal source pub­li­ca­tion: Branco, Jr., T., I. M. Kawashita, F. de Sá-Soares and C. N. Mon­teiro (2020). An IoT Appli­ca­tion Case Study to Opti­mize Elec­tric­ity Con­sump­tion in the Gov­ern­ment Sec­tor. Pro­ceed­ings of the Inter­na­tional Con­fer­ence on Elec­tronic Gov­ern­ment 2020—ICE­GOV 2020. Athens (Greece).
The final pub­li­ca­tion is avail­able here.

An IoT Appli­ca­tion Case Study to Opti­mize Elec­tric­ity Con­sump­tion in the Gov­ern­ment Sec­tor

Teó­filo T. Branco Júnior,^a Ilka M. Kawashita,^a Fil­ipe de Sá-Soares^a and Cláu­dio N. Mon­teiro^b

^a Algo­ritmi Cen­ter, Uni­ver­sity of Minho Guimarães, Por­tu­gal
^b Mod­u­lus One Bahia, Brazil

1. Introduction

Presently, the con­trol of elec­tric­ity con­sump­tion in Sal­vador City Hall’s facil­i­ties is man­u­ally car­ried out that gen­er­ates high oper­at­ing costs and causes unnec­es­sary expenses due to waste. Fre­quently, abnor­mal con­sump­tion caused by a defect is noticed after the fact and is only detected because it had caused an abrupt increase in energy con­sump­tion.

The con­cern with the con­sump­tion of elec­tric­ity resources and the mit­i­ga­tion of prob­lems related to its use are part of an insti­tu­tional effort to adopt smart gov­ern­ment (smart­gov) and smart cities admin­is­tra­tion-related poli­cies to pro­mote the best man­age­ment, gov­er­nance, and sus­tain­abil­ity prac­tices. This study is part of an ini­tia­tive of the Sec­re­tariat of Inno­va­tion and Sus­tain­abil­ity (SECIS), which in col­lab­o­ra­tion with the munic­i­pal tech­nol­ogy com­pany, the Elec­tronic Gov­er­nance Com­pany (COGEL), pre­pares stud­ies in part­ner­ship with cen­ters of tech­no­log­i­cal excel­lence and the aca­d­e­mic sci­en­tific com­mu­nity.

This study aimed to inves­ti­gate the effec­tive­ness of con­trol sys­tems based on IoT already used in the pri­vate sec­tor to con­trol the use of nat­ural resources in the pub­lic sec­tor. Focus­ing on iden­tify poten­tial gains and the scal­a­bil­ity of these sys­tems and orga­ni­za­tional behav­iours required for the imple­men­ta­tion.

This paper is orga­nized as fol­lows. In the intro­duc­tion, we present the research objec­tives and the method­ol­ogy that guided our stud­ies. In Sec­tion 2, we describe the the­o­ret­i­cal approach, while sec­tion 3 describes the site where this study was con­ducted. Sec­tion 4 describes the tech­no­log­i­cal archi­tec­ture, i.e. the char­ac­ter­is­tics of the applied tech­no­log­i­cal solu­tion, the IoT plat­form, sen­sors’ tech­ni­cal char­ac­ter­is­tics, the net­work topol­ogy, the spec­i­fi­ca­tion of col­lected data, and the soft­ware used for mon­i­tor­ing and con­trol­ling the sen­sor mod­ules. In sec­tion 5, we present the mea­sure­ments’ results and their respec­tive ana­lyzes. In sec­tion 6, we dis­cuss the results’ pro­jec­tions and the expected finan­cial and eco­nomic impact of the adop­tion of this solu­tion in all facil­i­ties of Sal­vador City Hall and present our con­clu­sion. We also rec­om­mend some behav­ioral prac­tices when adopt­ing this solu­tion through­out Munic­i­pal pub­lic admin­is­tra­tion. Finally, we present paths for future research.

1.1 Research Objectives

Senior man­age­ment of Sal­vador City Hall noticed that munic­i­pal gov­ern­ment pub­lic agen­cies had con­sumed high lev­els of elec­tric­ity in 2018. This fact trig­gered a series of stud­ies that led munic­i­pal pub­lic man­agers to inves­ti­gate and pro­pose solu­tions for improv­ing elec­tric­ity con­sump­tion. These stud­ies were con­ducted from June to Sep­tem­ber 2019.

In this sense, the pur­pose of the stud­ies was to ana­lyze and pro­pose a tech­no­log­i­cal solu­tion to improve the use of elec­tric energy resources in pub­lic munic­i­pal facil­i­ties. The solu­tion mon­i­tors and con­trols elec­tric­ity con­sump­tion using low-cost elec­tronic sens­ing devices, read­ily avail­able in the mar­ket, which are man­aged remotely in an auto­mated way. Smart pan­els show con­sump­tion infor­ma­tion. Pan­els show con­sump­tion graphs and sta­tis­tics over any period, allow sched­ul­ing alerts in the sys­tem to auto­mat­i­cally detect non-stan­dard occur­rences such as unex­pected power con­sump­tion caused by insta­bil­ity of the sup­ply sys­tem or defec­tive equip­ment. Smart pan­els allow the admin­is­tra­tor to proac­tively inter­vene in the power sup­ply, in case of detec­tion of any anom­aly that may affect sup­ply. He can request peri­odic main­te­nance for the net­work and equip­ment or can estab­lish guide­lines for the use of air con­di­tion­ers based on con­sump­tion pro­files.

MOD­U­LUS ONE is a local pri­vate com­pany with exper­tise in inno­v­a­tive engi­neer­ing and tech­nol­ogy solu­tions involv­ing IoT. It pro­vided the tech­no­log­i­cal solu­tion, hard­ware and soft­ware, and the stud­ies for solu­tion’s suit­abil­ity for the gov­ern­ment sec­tor were per­formed by tech­nol­ogy ana­lysts and tech­ni­cians of the same com­pany.

This research aims to achieve the fol­low­ing results:

1. Iden­tify elec­tric­ity con­sump­tion pro­files in a given pub­lic envi­ron­ment;

2. Ver­ify if the analy­sis of the mea­sure­ment con­trols can indi­cate ways to improve elec­tric­ity use;

3. Eval­u­ate if the pro­posed solu­tions can be applied to other munic­i­pal admin­is­tra­tion facil­i­ties.

We expect that the ana­lyzes of this study can be applied to all gov­ern­ment agen­cies oper­at­ing in this munic­i­pal­ity.

1.2 Methodology

Cur­rent research explores the pub­lic admin­is­tra­tion’s per­cep­tion of elec­tric­ity sav­ings in pub­lic build­ings and pro­vides evi­dence of the accep­tance of the use of smart tech­nolo­gies for pub­lic admin­is­tra­tion mon­i­tor­ing and con­trol of this resource.

The study was per­formed based on quan­ti­ta­tive mea­sure­ments of data col­lected in an IoT data­base and ana­lyzed using algo­rithms coded into the con­trol soft­ware. The tech­ni­cal stan­dard PRODIST 8 [ANEEL 2018] was used to detect devi­a­tions of elec­tri­cal volt­age and to peri­od­i­cally mea­sure the con­sump­tion recorded for the var­i­ous sen­sors installed on the grid.

The con­sump­tion results pre­sented by the algo­rithm embed­ded in the con­trol soft­ware allow total­ing the con­sump­tions for each air con­di­tioner in any pre­de­ter­mined unit of time. In order to iden­tify the con­sump­tion in dif­fer­ent envi­ron­ments and dif­fer­ent equip­ment in one facil­ity, the cal­cu­lated totals are com­pared against the observed val­ues for a given time frame. Thus, it is pos­si­ble to iden­tify the peri­ods when con­sump­tion is high. A lin­ear regres­sion analy­sis [Klein­baum and Kup­per 1978] was used to iden­tify the equip­ment con­sump­tion in a time unit of ref­er­ence.

Root cause analy­sis (RCA) is designed for inves­ti­gat­ing and cat­e­go­riz­ing the root causes of events with safety, health, envi­ron­men­tal, qual­ity, reli­a­bil­ity and pro­duc­tion impacts [Rooney and Van­den Hau­vel 2004]. An RCA, through its cause-and-effect dia­gram, was per­formed to iden­tify con­sump­tion devi­a­tions and for­ward pos­i­tive rec­om­men­da­tions to improve user’s energy uti­liza­tion.

2. Background

The tech­no­log­i­cal evo­lu­tion achieved in recent times pro­vides con­di­tions for the cre­ation of solu­tions involv­ing inno­v­a­tive con­cepts and has the poten­tial to be applied in the gov­ern­ment sec­tor pro­mot­ing sus­tain­abil­ity. These solu­tions help improve oper­a­tional effi­ciency, and when cou­pled with process opti­miza­tion can reduce the use of nat­ural resources. The solu­tion pro­posed explores con­cepts and tech­nolo­gies such as e-Gov­ern­ment, Smart Gov­ern­ments, Smart Tech­nolo­gies, and IoT for improv­ing and opti­miz­ing gov­ern­ment oper­a­tions man­age­ment. In order to bet­ter under­stand the poten­tial offered by these tech­no­log­i­cal and inno­v­a­tive con­cepts, we sought to iden­tify some works that address these con­cepts and report expe­ri­ences sim­i­lar to this study. This frame­work aimed to clar­ify impor­tant details in the appli­ca­tion of these tech­nolo­gies and gather expe­ri­ences of lessons learned in other loca­tions.

2.1 e-Government

Accord­ing to Seifert [2003], e-gov­ern­ment aims to con­tin­u­ously opti­mize ser­vice deliv­ery, social par­tic­i­pa­tion, and gov­er­nance by trans­form­ing inter­nal and exter­nal rela­tion­ships through tech­nol­ogy, the Inter­net, and new media. The author states that e-gov­ern­ment has six cat­e­gories: Gov­ern­ment that pro­vides ser­vices to indi­vid­u­als (G2IS), Gov­ern­ment to indi­vid­u­als as part of the polit­i­cal process (G2IP), Gov­ern­ment to busi­nesses as a cit­i­zen (G2BC), Gov­ern­ment to busi­nesses in the mar­ket (G2BMKT), gov­ern­ment to employ­ees (G2E) and gov­ern­ment to gov­ern­ment (G2G).

For Omura [2000], among many other advan­tages, e-gov­ern­ment will lead to a reduc­tion in paper con­sump­tion for gov­ern­ment-busi­ness trans­ac­tions (G to B trans­ac­tions).

These tech­nolo­gies can serve a vari­ety of dif­fer­ent pur­poses: bet­ter deliv­ery of gov­ern­ment ser­vices to cit­i­zens, bet­ter inter­ac­tions with busi­nesses and indus­tries, cit­i­zen empow­er­ment through access to infor­ma­tion, or more effi­cient gov­ern­ment man­age­ment. The result­ing ben­e­fits may be less cor­rup­tion, greater trans­parency, greater con­ve­nience, rev­enue growth, and cost sav­ings [Almarabeh et al. 2016].

The rela­tion­ship between gov­ern­ment and a pri­vate com­pany pro­vid­ing auto­mated energy man­age­ment solu­tion ser­vices con­fig­ures a rela­tion­ship of the G2BMKT cat­e­gory. In this sce­nario, the gov­ern­ment can inter­act with the com­pany, acquir­ing solu­tions for its admin­is­tra­tive processes, seek­ing oper­a­tion costs reduc­tion.

2.2 Smart Government and Smart Technologies

An intel­li­gent city exploits sus­tain­able Infor­ma­tion and Com­mu­ni­ca­tion Tech­nolo­gies (ICT) to improve the qual­ity and per­for­mance of urban ser­vices for cit­i­zens and gov­ern­ment while reduc­ing resource con­sump­tion [Kha­je­nasiri et al. 2007].

Schu­ur­man and Tõnurist [2017] define smart gov­er­nance as the process of col­lect­ing all types of data and infor­ma­tion related to sen­sors or sen­sor net­works pub­lic man­age­ment. New tech­nolo­gies are har­nessed to strengthen gov­ern­ment ratio­nale by using a com­plete - and more read­ily avail­able and acces­si­ble - infor­ma­tion for gov­ern­ment deci­sion-mak­ing processes and the imple­men­ta­tion of those deci­sions.

Smart gov­ern­ment is used to char­ac­ter­ize activ­i­ties that cre­atively invest in emerg­ing tech­nolo­gies, along with inno­v­a­tive strate­gies to achieve more agile and resilient gov­ern­ment struc­tures and gov­er­nance infra­struc­tures [Gil-Gar­cia et al. 2014]. The authors argue that an “intel­li­gent state” is a new form of elec­tronic gov­er­nance that uses sophis­ti­cated infor­ma­tion tech­nolo­gies to inter­con­nect and inte­grate infor­ma­tion, processes, insti­tu­tions and phys­i­cal infra­struc­ture to serve cit­i­zens and com­mu­ni­ties bet­ter. This kind of smart gov­er­nance is at a higher level of trans­for­ma­tion because it requires restruc­tur­ing the inter­nal orga­ni­za­tion of gov­ern­ment: admin­is­tra­tions need to be inno­v­a­tive in address­ing dif­fer­en­ti­ated pol­icy require­ments.

Accord­ing to Zheng et al. [2013], the smart meter is one of the most impor­tant devices used in the smart grid. The smart meter is an advanced power meter that gath­ers infor­ma­tion from end-user charg­ing devices and mea­sures con­sumer power con­sump­tion and then pro­vides addi­tional infor­ma­tion to the util­ity or sys­tem oper­a­tor. Var­i­ous sen­sors and con­trol devices, sup­ported by ded­i­cated com­mu­ni­ca­tion infra­struc­ture, are used in one smart meter.

2.3 Internet of Thinks (IoT)

Inter­net of Things (IoT) is a net­work that allows new forms of com­mu­ni­ca­tion between peo­ple and things and between things them­selves. Each of the things or objects in IoT com­mu­ni­cates with each other and plays a spe­cific role. In an IoT net­work, each node acquires infor­ma­tion on their own, and humans only ver­ify the infor­ma­tion col­lected [Park et al. 2014]. IoT can be used in trans­porta­tion, health­care, intel­li­gent envi­ron­ments, etc. The main net­work sys­tems for com­mu­ni­cat­ing with IoT are radio fre­quency iden­ti­fi­ca­tion (RFID) sys­tems, wire­less sen­sor net­work (WSN), and RFID sen­sor net­work (RSN) [Park et al. 2014].

In a world of multi-stake­holder infor­ma­tion and asset deliv­ery, and mil­lions of things inter­act­ing and com­mu­ni­cat­ing in real-time, IoT-based sys­tems aim to exploit these assets in a resilient and sus­tain­able man­ner [Kyr­i­azis et al. 2013].

Intel­li­gent energy con­trol in build­ings is an impor­tant aspect of that world. The Inter­net of Things can pro­vide a solu­tion. Its goal is to con­nect mul­ti­ple het­ero­ge­neous devices across the Inter­net, sup­ported by a flex­i­ble tiered archi­tec­ture, where things, peo­ple, and Cloud ser­vices are com­bined to make an appli­ca­tion task eas­ier. A smart home can be con­sid­ered a sub­cat­e­gory of smart cities. In this sub­cat­e­gory, house­hold appli­ances, light­ing, heat­ing, and air con­di­tion­ing sys­tems, video and audio devices, and secu­rity sys­tems are capa­ble of com­mu­ni­cat­ing with each other or use a cen­tral con­trol unit to com­mu­ni­cate, in order to pro­vide com­fort, safety, and energy effi­ciency for home­own­ers [Kha­je­nasiri et al. 2007].

2.4 Related Works

The use of IoT for mon­i­tor­ing of elec­tric­ity con­sump­tion is a well-known prac­tice, notably to homes and pri­vate build­ings. More­over, some authors report suc­cess­ful case expe­ri­ences in the pub­lic sec­tor regard­ing the use of sim­i­lar smart equip­ment.

Energy-sav­ing is a major issue due to the pro­lif­er­a­tion of cli­mate changes and global energy chal­lenges [Bhati et al. 2017]. Accord­ing to these authors, peo­ple’s per­cep­tion of the use of smart tech­nol­ogy for energy sav­ing is still in its con­cep­tual phase. It means that peo­ple talk about envi­ron­men­tal aware­ness, but do not change their con­sump­tion habits and con­tinue to pay the elec­tric­ity tab. Given the avail­abil­ity of elec­tric­ity and its essen­tial role, mod­u­lat­ing con­sumer atti­tudes toward energy sav­ings can be a chal­lenge. Also, accord­ing to Bhati et al. [2017], the cur­rent design gap of smart
tech­nol­ogy in smart homes is under­stand­ing con­sumer behav­ior and inte­grat­ing that under­stand­ing into smart tech­nol­ogy.

Exam­ples of smart cities often high­light San­tander, Spain, a test site for the world’s most “sen­sor-equipped” city. The city has installed over 20,000 fixed and mobile sen­sors to exam­ine park­ing trends, man­age power sup­plies for schools, build­ings, and street light­ing [Gil-Gar­cia et al. 2014].

Vas­sileva and Campillo [2016] report a case study of a large-scale imple­men­ta­tion of smart meters inte­grated into a smart power grid in the city of Vasteras, Swe­den. They con­cluded that the advan­tage of smart meters is the abil­ity to col­lect con­sumer data, which is eas­ily acces­si­ble through a web­site or mobile app. Con­sumers can overview of their con­sump­tion pat­terns, what they can act on, and how to reduce their use of devices. The sur­vey also revealed some gaps in con­sumer inter­ac­tions with smart meters, as can be seen from con­sumer feed­backs and pref­er­ences on how smart meters can be enhanced. Another advan­tage of smart meters was the dynamic use of tar­iff-based elec­tric­ity that allows con­sumers to be guided at peak times and in choos­ing spe­cific energy sup­pli­ers.

An addi­tional case study reported by Bhati et al. [2017] regarded the instal­la­tion of smart meters and online dis­play of house­hold elec­tric­ity con­sump­tion in Sin­ga­pore, one of the most densely pop­u­lated cities in Asia. Energy-sav­ing behav­iors in homes were inves­ti­gated using home dis­plays and smart meters. Raw energy data, sta­tis­ti­cal data, and back­ground infor­ma­tion were stored in ded­i­cated data­bases so that researchers could work on their data sep­a­rately. The find­ings of this study were cat­e­go­rized into top­ics such as fre­quency check­ing, power sav­ing, standby power sav­ing, and in-peak and off-peak power con­sump­tion. House­hold energy con­sump­tion was smoothly dis­trib­uted over 24 hours. In homes where there were no screens to mon­i­tor con­sump­tion, the demand for energy was much higher between 7 p.m. to 11 p.m. com­pared to fam­i­lies who could mon­i­tor their con­sump­tion. This case study demon­strated that house­holds’ energy-sav­ing behav­ior had changed pos­i­tively due to the instal­la­tion of smart home tech­nolo­gies.

Park et al. [2014] present another study of the appli­ca­tion of IoT tech­nol­ogy using elec­tric­ity mea­sure­ment sen­sors in his­tor­i­cal build­ings in the city of Sin­ga­pore. The authors were involved in the devel­op­ment of an ICT-based power man­age­ment con­trol sys­tem to pre­vent pos­si­ble dam­age caused by major facil­i­ties inter­ven­tions due to the instal­la­tion of power man­age­ment hard­ware. In this project, a com­puter-based sys­tem con­trols light­ing, heaters, air con­di­tion­ers, and other envi­ron­men­tal units in large build­ings. This solu­tion pro­vides a soft­ware-level cen­tral con­trol sys­tem that con­nects wire­lessly to power struc­tures placed in dif­fer­ent parts of a build­ing or even in mul­ti­ple build­ings. In addi­tion to the hard­ware, the most sig­nif­i­cant result was the devel­op­ment of an energy effi­ciency model for exist­ing build­ings and pub­lic spaces. These sys­tems use IoT soft­ware and hard­ware infra­struc­tures to achieve their goals.

3. Study Site

The place cho­sen for pilot­ing the exper­i­ment was the COLA­BORE cen­ter for tech­no­log­i­cal inno­va­tion, spe­cially designed by Sal­vador City Hall to test new smart and sus­tain­able tech­nolo­gies for the pub­lic sec­tor. Fig­ure 1 shows a pic­ture of the inno­va­tion cen­ter facil­i­ties located in one of the city’s munic­i­pal wooded parks.

Fig­ure 1: COLA­BORE Inno­va­tion Cen­ter
Source: Sal­vador City Hall

In this space, recy­cling dis­used ship con­tain­ers, the city built eight inte­grated build­ings. These con­tain­ers were designed to serve as a col­lec­tive space for offices and class­rooms. Gov­ern­ment, pri­vate enti­ties, and star­tups focused on inno­va­tion, tech­nol­ogy, and sus­tain­abil­ity share the area to pro­mote entre­pre­neur­ship ini­tia­tives. Some spaces are also avail­able for pub­lic use. Any inter­ested party can book a space to per­form var­i­ous activ­i­ties.

All build­ings have lawns cov­er­ing their roofs and a plumb­ing sys­tem to col­lect and reuse rain­wa­ter and water that drip from air con­di­tion­ing (AC) units. The com­plex also has a small sewage treat­ment plant which, by pump­ing, stores it in tanks located on the top of the con­tain­ers. Waters are used to irri­gat­ing the lawn on the roof of the con­tain­ers and sur­round­ing green areas and recy­cled in sinks and toi­lets.

This space also has other inno­v­a­tive ini­tia­tives, such as pho­to­elec­tric cells pan­els for har­ness­ing sun­light, acces­si­bil­ity solu­tions, and many oth­ers. More recently, the sen­sor mod­ules, object of this study, were installed in these facil­i­ties. The net­work infra­struc­ture and the mod­ules con­tain­ing the sen­sors were installed in June 2019. Mon­i­tor­ing for this study occurred until Sep­tem­ber 2019.

The inno­va­tion cen­ter is com­posed of 8 mar­itime con­tain­ers. Six con­tain­ers have two air con­di­tion­ing devices; one con­tainer has three units; and one with one air con­di­tioner. Six­teen sen­sor mod­ules were installed, one for each equip­ment. The instal­la­tion process was straight­for­ward since the mod­ules work in plug & play mode, and no con­fig­u­ra­tion was required.

The instal­la­tion of the sen­sors was sim­ple and car­ried out in one day. As the elec­tri­cal cir­cuits that power each air con­di­tion­ing (AC) unit are insu­lated to pre­vent short cir­cuits, it was pos­si­ble to install each sen­sor mod­ule besides the AC unit. The energy that feeds each equip­ment passes first through the mod­ule and enables the mod­ule’s cut-off relay to take con­trol of the energy flow to allow the unit to oper­ate accord­ing to pre­vi­ously pro­grammed para­me­ter­i­za­tion.

4. Technological Architecture

The sys­tem archi­tec­ture is com­posed of the fol­low­ing infra­struc­ture: mod­ules con­tain­ing the read­ing sen­sors/actu­a­tors (meters), a gate­way that is used as a bridge, the net­work infra­struc­ture, the Cloud com­put­ing, a data­base that sup­ports BIG­DATA, and the soft­ware for sen­sors’ remote man­age­ment. Fig­ure 2 presents the tech­no­log­i­cal archi­tec­ture used in this study.

Fig­ure 2: Tech­no­log­i­cal Archi­tec­ture

All devices of this plat­form, described below, are enabled with IoT tech­nol­ogy.

4.1 Sensor Modules

The sen­sor mod­ules are respon­si­ble for receiv­ing and send­ing the elec­tric­ity flow to the remote man­age­ment sys­tem. Mod­ules can act as actu­at­ing devices, inter­rupt­ing the elec­tric­ity flow to the air con­di­tion­ers accord­ing to pre­pro­grammed para­me­ters.

The sen­sor mod­ules pass the energy flow infor­ma­tion to a Gate­way, allow­ing power teleme­try mea­sure­ments and event alarms to reach the mod­ules’ remote man­age­ment sys­tem.

Each mod­ule has Mesh net­work func­tion­al­ity and oper­ates via radiofre­quency. It is pow­ered by a cur­rent which volt­age ranges from 90 to 250 Vac. In standby, con­sump­tion is set to 1.5 W and assumes the nom­i­nal con­sump­tion of the asso­ci­ated AC unit when in oper­a­tion. The radiofre­quency mod­ule with phys­i­cal layer FSK mod­u­lated has a zero-gain antenna for receiv­ing sig­nals and works in an ISM band on 918 MHz fre­quency. The mod­ule oper­ates in the fre­quency range from 915 to 928 MHz.

Each mod­ule is attached to the fol­low­ing periph­eral devices: a shunt relay, a real-time clock, a tem­per­a­ture sen­sor onboard the moth­er­board for mea­sur­ing the mod­ule’s inter­nal tem­per­a­ture, a power meter for mea­sur­ing elec­tri­cal quan­ti­ties such as the elec­tric cur­rent, volt­age, active and reac­tive power, and one sen­sor to detect unau­tho­rized inter­ven­tions. This sys­tem mea­sures the air con­di­tioner per­for­mance and detects degra­da­tion, in which case it issues event alerts that indi­cate the need for main­te­nance inter­ven­tion.

It is pos­si­ble to make these mea­sure­ments of each of the air con­di­tion­ers indi­vid­u­ally or grouped in a pre-estab­lished set. The sen­sor mod­ule can detect power sup­ply anom­alies that can cause some unit oper­a­tion prob­lems.

The real-time clock works with energy stored in a super capac­i­tor and does not require bat­ter­ies. The micro­con­troller allows the mod­ule to pro­gram time pro­files to deter­mine the acti­va­tion and shut­down times of an air con­di­tion­ing unit. The mod­ule does not per­form the tem­per­a­ture mod­u­la­tion sys­tem. It is oper­ated directly on the ther­mo­stat of the air con­di­tioner, using the orig­i­nal con­trol of the unit.

The pro­pri­etary com­mu­ni­ca­tion pro­to­col is devel­oped in open source and is capa­ble of inte­grat­ing other mod­ules through APIs. (OMIT­TED ON PUR­POSE), the devices’ sup­plier, per­formed all pro­gram­ming and devel­oped and imple­mented all net­work link layer pro­to­cols.

4.2 Gateway Bridge

A Gate­way Bridge device has been installed in the COLA­BORE space. This bridge acts as a com­mu­ni­ca­tion link between the field net­work that oper­ates on radio fre­quency and the Inter­net net­work that trans­ports the col­lected data to the remote man­age­ment plat­form.

Each bridge can receive and send infor­ma­tion for up to 1000 sen­sor mod­ules. The device runs LINUX soft­ware. A bridge has embed­ded com­mu­ni­ca­tion soft­ware, which enables sen­sor mod­ules to com­mu­ni­cate via radiofre­quency. Data com­mu­ni­ca­tion to a Cloud Com­put­ing is made pos­si­ble through the Inter­net net­work. It is desir­able to have a sta­ble Inter­net net­work with appro­pri­ate band­width to avoid time-out effects when sig­nals are expired by the Inter­net net­work low latency. Such effects, how­ever, do not inter­fere with the oper­a­tion of sen­sor mod­ules, which oper­ate through self-stored pro­grams.

This device oper­ates with elec­tri­cal volt­age in the range of 90 to 250 Vac. Using a pub­lic Inter­net IP, the bridge com­mu­ni­cates with the sys­tem appli­ca­tion servers on the Cloud. The device can con­nect to the Inter­net using the local Inter­net net­work or oper­at­ing with 3G or 4G tech­nolo­gies or using the met­ro­pol­i­tan mobile net­work. Data always trav­els through VPN tun­nel­ing and uses AES128 / 256 stan­dard encryp­tion. The sen­sor mod­ules solve any speed prob­lems through bound­ary pro­cess­ing. In the bound­ary process, field devices per­form analy­sis and actu­a­tion on the sen­sor mod­ule’s own pro­ces­sor, reduc­ing gate­way pro­cess­ing and band­width con­sump­tion.

The dif­fer­ence between this Gate­way and oth­ers on the mar­ket­place is that it does not assume the role of net­work coor­di­na­tor, in which case the device stores routes and pairs com­mu­ni­ca­tion with other mod­ules. In this case, if the device mal­func­tions, all pro­gram­ming is lost.

The Gate­way device used in this solu­tion works with a dis­trib­uted intel­li­gence par­a­digm, i.e., the sen­sor mod­ules do the bound­ary pro­cess­ing and send the data to the other mod­ules of the sys­tem, oper­at­ing on Mesh net­work prin­ci­ples. Whereas in the dis­trib­uted pro­gram­ming par­a­digm, sen­sor mod­ules can record pro­gram­ming and the gate­way does not assume the role of coor­di­na­tor, per­form­ing only as a bridge of com­mu­ni­ca­tion with the Inter­net net­work through a VPN. There is no pair­ing between the gate­way and the sen­sors. Pair­ing is done through the sen­sors them­selves.

In case of fail­ure, the Gate­way can be replaced with a sim­i­lar device, pre­serv­ing only the pub­lic IP. And the nodes of the net­work are not affected; thus, the sen­sor mod­ules pre­serve all their pro­gram­ming and set­tings.

4.3 Mesh Network

The com­mu­ni­ca­tion net­work infra­struc­ture between the sen­sor mod­ules and the gate­way is through a mesh net­work.
The wire­less Mesh net­work is based on the 802.11 wire­less local area net­work (WLAN) and has been actively used for some years. In order to improve the per­for­mance of WLAN mesh net­works, some new com­mu­ni­ca­tion pro­to­cols have been devel­oped in recent years [Wang and Lim 2008]. Fig­ure 3 shows a schematic of the 802.11s archi­tec­ture Mesh net­work.

The sys­tem’s Mesh net­work infra­struc­ture is com­posed of the set of com­mu­ni­ca­tions nodes and the bridge. In this type of net­work, all nodes com­mu­ni­cate directly to each other. Com­mu­ni­ca­tion does not require rout­ing tables. This con­fig­u­ra­tion par­a­digm makes installing mod­ules and gate­ways eas­ier, result­ing in greater resilience in over­com­ing obsta­cles or shad­ows. For exam­ple, when the gate­way sends a sig­nal one km away, and there are sen­sors along the way to pick up this sig­nal, the net­work nodes will repeat the sig­nal to the other nodes. Nodes that belong to dif­fer­ent groups can trans­mit other types of sig­nals in dif­fer­ent groups, so the sig­nal can reach the gate­way.

Fig­ure 3: Mesh Net­work
Adapted from [Wang and Lim 2008, p. 972]

The Mesh net­work envi­ron­ment has the fol­low­ing char­ac­ter­is­tics: use of IoT data com­mu­ni­ca­tion pro­to­col, applic­a­ble to RF Mesh net­works and based on open tech­nolo­gies; dis­trib­uted intel­li­gence - no coor­di­nat­ing ele­ments; no rout­ing tables (com­mu­ni­ca­tion resiliency); inte­gra­tion of dif­fer­ent devices form­ing a het­ero­ge­neous net­work; def­i­n­i­tion of all lay­ers; sin­gle-com­mand com­mis­sion­ing devices

Com­mu­ni­ca­tion pro­to­cols pos­sess the fol­low­ing tech­ni­cal char­ac­ter­is­tics: Gate­way plug & play; AES128/256 encryp­tion; FSK and LoRa mod­u­la­tions; and Inte­gra­tion API.

The min­i­mum require­ments for infra­struc­ture oper­a­tion involve a server with 64-bit dual-core Linux vir­tual machine, 30 Gb hard disk and 4 Gb of RAM.

Although Mesh com­mu­ni­ca­tions can travel over a reg­u­lar net­work, sig­nals from dif­fer­ent devices can be grouped so that they are not shared. This process pro­vides greater com­mu­ni­ca­tion resilience for field devices. The pro­gram­ming of the sig­nal trans­mis­sion code does not allow the sys­tem to go into infi­nite loops, which pre­vent net­work con­ges­tion.

4.4 Cloud Computing

Cloud com­put­ing is loca­tion-inde­pen­dent com­put­ing where shared servers pro­vide resources, soft­ware, and data to com­put­ers and other devices on demand. More sim­ply, it can be seen as remote com­put­ing [Ramya and Ramya 2015]. Fig­ure 4 shows a schematic of a Cloud.

The Cloud com­put­ing infra­struc­ture adopted for this solu­tion was Ama­zon’s Plat­form as a Ser­vice (PaaS) AWS-EC2 [AWS 2019].
Servers are housed in two data cen­ters: one in South Amer­ica, located in Sao Paulo, Brazil and the sec­ond in Europe, located in Ire­land. Data cen­ters have been con­tracted with load bal­anc­ing and allow pro­cess­ing and data mir­ror­ing.

Fig­ure 4: Cloud Com­put­ing
Source: [Ramya and Ramya 2015, p. 1]

The sys­tem uses a region­al­ized DNS depend­ing on the loca­tion of the server that accesses the sys­tem that con­nects the near­est Cloud server. In the event one server is unavail­able, the other takes over data stor­age. Stor­age ref­er­ence is per­formed by geo­graphic ref­er­enc­ing. The Cloud iden­ti­fies the con­nect­ing gate­way using its IP geo­graph­i­cal loca­tion and directs the traf­fic to the appro­pri­ate data cen­ter.

4.5 NoSQL Database MongoDB

Power flow data cap­tured from field sen­sor mod­ules pass through the plat­form Gate­way bridge to be stored in a NoSQL Mon­goDB Cloud data­base.

Sen­sors are used to mon­i­tor cer­tain aspects of the phys­i­cal or vir­tual world, and data­bases are often used to store the data these sen­sors gather. Sen­sor usage is increas­ing, which leads to increas­ing demand for sen­sor data stor­age plat­forms. Some sen­sor mon­i­tor­ing appli­ca­tions need to auto­mat­i­cally add new data­bases as the size of the sen­sor net­work grows. In this con­text, NoSQL data­bases have gained strength over the past two years due to the increased scal­a­bil­ity and avail­abil­ity require­ments [Poko­rny 2013].
NoSQL data­bases are designed for sim­ple hard­ware. In addi­tion to cloud com­put­ing, NoSQL data­bases are pop­u­lar among Web 2.0 appli­ca­tions, where the hor­i­zon­tal scale involves thou­sands of nodes. It is no coin­ci­dence that NoSQL data­bases with the great­est impact on this soft­ware cat­e­gory orig­i­nate from Google and Ama­zon devel­op­ment labs [Poko­rny 2013].

Mon­goDB is a doc­u­ment-ori­ented, non­re­la­tional data­base. Appli­ca­tion dataset sizes are grow­ing at an incred­i­ble rate. Increas­ing avail­able band­width and cheap stor­age have cre­ated an envi­ron­ment in which even small-scale appli­ca­tions need to store more data than many data­bases can store [Chodorow 2019].

4.6 Remote Management Software

The soft­ware plat­form for remote man­age­ment was devel­oped by the com­pany MOD­U­LUS to process the data mea­sure­ments made by the sen­sor mod­ules.

The soft­ware com­bines sev­eral pro­gram­ming lan­guages. One part uses HTML5, Java, and Python. It uses ERLANG, a tele­com­mu­ni­ca­tion spe­cific lan­guage, for some func­tions backed com­mu­ni­ca­tion.

Within this plat­form, users are allowed to view data in graphs or tables. Data can be exported to for­mats that may be con­sum­able for other BIG­DATA tech­nolo­gies for fur­ther analy­sis.

The plat­form allows to view teleme­try, cre­ate sta­tis­tics, and view elec­tri­cal devices power con­sump­tion, pro­gram­ming oper­a­tion para­me­ters, although the code is stored in the net­work nodes.

The sys­tem per­mis­sions mod­ule allows the def­i­n­i­tion of var­i­ous user access lev­els. Spe­cific per­mis­sions can be set up for the admin­is­tra­tor, edi­tor, or viewer cat­e­gories. The admin­is­tra­tor can per­form all oper­a­tions avail­able on the sys­tem. The edi­tor role has access to some set­tings but can­not reg­is­ter new users. The viewer role can only query data. Viewer pro­files can be spe­cific to restricted or unre­stricted viewer func­tions. Pro­files allow con­fig­ur­ing net­works where view­ers are given access to resources through per­mis­sions assign­ment. MOD­U­LUS owns the soft­ware and makes it avail­able to users under an annual license fee.

5. Results

COLA­BORE inno­va­tion cen­ter ana­lyzed the teleme­try results, which con­sist of data of one facil­ity com­posed of eight envi­ron­ments with 16 air con­di­tion­ing units.

First, we deter­mined the con­sump­tion of the air con­di­tion­ing sys­tem dur­ing the entire test­ing period from June to Sep­tem­ber 2019. Then we com­pared it to the total con­sump­tion of the gov­ern­ment facil­ity to assess the rep­re­sen­ta­tive­ness of the air con­di­tion­ers in regard to the total energy con­sump­tion. Sec­ondly, we restricted the study period to the week 2019/09/14 to 2019/09/22 to iden­tify the daily con­sump­tion of air con­di­tion­ing units and learn users’ behav­iors habits.

For this exper­i­ment, sen­sor mod­ule para­me­ters were not ini­tially pro­grammed, so sen­sors could per­form unbounded. The inten­tion was to cre­ate a base­line with teleme­try of each air con­di­tion­ing unit actual elec­tric­ity con­sump­tion and observe how users behave, in order to iden­tify their usage habits.

This study obtained the fol­low­ing results:

1. Total con­sump­tion for the test period from 2019/06/22 to 2019/09/22—One of the objec­tives of this study was to deter­mine the rela­tion between the power con­sump­tion of all air con­di­tion­ing units and the facil­ity’s total elec­tric­ity con­sump­tion.

2. Teleme­try data analy­sis—Occur­rences in con­sump­tion of the use of air con­di­tion­ing units were iden­ti­fied. The demands of using the devices in their respec­tive envi­ron­ments were mea­sured. The analy­sis of these data allowed us to estab­lish spe­cific usage pro­files for these envi­ron­ments.

3. Sim­u­la­tions—Sim­u­la­tions of use were per­formed in an envi­ron­ment con­trolled by auto­mated pro­grams ver­sus mea­sures acquired for the base­line (unre­stricted use). The ana­lyzes allowed to ver­ify if the con­trolled envi­ron­ment can point out ways to improve elec­tric­ity con­sump­tion;

4. Behav­ior iden­ti­fi­ca­tion—User behav­iors were iden­ti­fied, and the chal­lenges for ratio­nal and opti­mized use of the sys­tem were listed.

5.1 Test Period Total Energy Consumption

This study mea­sured the energy con­sump­tion of the air con­di­tion­ers from 2019/06/22 to 2019/09/22. Dur­ing this period, air con­di­tion­ers con­sump­tion mea­sure­ment was 6.000,491 kWh. At the same period, the facil­ity’s elec­tric­ity bill indi­cated a total elec­tric­ity con­sump­tion of 23.094 kWh. There­fore, the elec­tric­ity con­sump­tion of all air con­di­tion­ing units cor­re­sponded to approx­i­mately 25.98% of the total con­sump­tion. Fig­ure 5 presents a graph rep­re­sent­ing the rela­tion between the elec­tric­ity con­sump­tion of air con­di­tion­ing units in regard to the facil­ity total power con­sump­tion.

Fig­ure 5: Con­sump­tion of Air Con­di­tion­ing Units

From June to Sep­tem­ber 2019, the Col­lab­o­rate Cen­ter spent approx­i­mately US$ 3,375, includ­ing taxes, on elec­tric­ity. The cost of the kW/h applied to Sal­vador City Hall is US$ 0.15. Thus, the energy con­sump­tion of the air con­di­tion­ing sys­tem was approx­i­mately US$ 900.

The rela­tion between the facil­ity’s total elec­tric­ity con­sump­tion and the air con­di­tion­ers has shown that the units are respon­si­ble of over a quar­ter of the total energy con­sump­tion.

This find­ing attests to the impor­tance of con­trol­ling the use of the air con­di­tion­ing units.

5.2 Telemetry Data and Analysis

The remote man­age­ment soft­ware pro­vided the teleme­try data and their analy­sis. To iden­tify users’ behav­iors and con­sump­tion habits related to the usage of air con­di­tion­ing units, we cre­ated a sam­ple of the results of a sys­tem query ran between 2019/09/14 and 2019/09/22.

This sam­ple is suf­fi­cient to eval­u­ate the poten­tial mon­i­tor­ing, man­age­ment, and con­trol imple­men­ta­tion ren­dered pos­si­ble by the use of this sys­tem. Thus, the sys­tem query mod­ule pro­vided the fol­low­ing sta­tis­tics dur­ing this period:

1. The actual state of sen­sor mod­ules - Pie chart con­tain­ing the per­cent­age of devices oper­at­ing nor­mally rel­a­tive to devices with non­con­form­ing events and the ration of events occur­rences/sen­sor.

2. The power con­sump­tion of the all air con­di­tion­ers in the facil­ity, in each envi­ron­ment per period (set by the sys­tem oper­a­tor). This sta­tis­tic is avail­able to the admin­is­tra­tor as fol­lows:

   1. a data table con­tain­ing con­sump­tion val­ues;

   2. a bar graph com­par­ing units’ con­sump­tion;

   3. a line graph indi­cat­ing daily con­sump­tion.

5.2.1 Sensor Module Actual State Chart

The pie chart indi­cates which sen­sors are in nor­mal oper­a­tion and which are not. Fig­ure 6 illus­trates the chart gen­er­ated by the sys­tem.

Fig­ure 6: Con­trol Chart

On 2019/09/22, the remote man­age­ment soft­ware pointed out that of the 16 sen­sor mod­ules, 14 were in nor­mal oper­a­tion, and two were defec­tive. The list of defec­tive mod­ules on this date is pre­sented in Table 1.

Table 1: Events

Analy­sis of these data indi­cated that there was evi­dence of man­ual han­dling of the sen­sor mod­ules, desta­bi­liz­ing them from their orig­i­nal posi­tions. The admin­is­tra­tor should request main­te­nance per­son­nel to check the sta­tus of these devices and to repro­gram their posi­tion in the sys­tem.

5.2.2. Electricity consumption of all air conditioning units

Table 2 presents the sta­tis­tics gen­er­ated by the sys­tem on the elec­tric­ity con­sump­tion of all eight envi­ron­ments of the facil­ity from 2019/09/14 to 2019/09/22.

Table 2: Total Air Con­di­tion­ing Elec­tric­ity Con­sump­tion in a Pub­lic Agency

Fig­ure 7 shows the con­sump­tion of all air con­di­tion­ing units for the period between 2019/09/14 and 2019/09/22.

Fig­ure 7: Con­sump­tion Read­ings

The analy­sis of the energy con­sump­tion in this facil­ity revealed the exis­tence of expres­sive dif­fer­ences of con­sump­tion between the air con­di­tion­ing units. The admin­is­tra­tor should inves­ti­gate why there are such dif­fer­ences in con­sump­tion to assess whether any admin­is­tra­tive action should be taken.

Fig­ure 8 presents a graph illus­trat­ing the total gov­ern­ment facil­ity’s elec­tric­ity con­sump­tion per day.
The analy­sis of these data pro­vides the admin­is­tra­tor means to deter­mine and set an expected con­sump­tion value of the facil­ity for a cer­tain period.

Fig­ure 8: Total Con­sump­tion per Day

Data observed in nor­mal sit­u­a­tions may sup­port the man­ager’s deci­sion to set a cap on energy con­sump­tion over sim­i­lar peri­ods.

Thus, event alarms can be pro­grammed to indi­cate con­sump­tion that exceeds the upper limit set by the admin­is­tra­tor.

5.2.3 Energy Consumption in a Specific Government Facility Environment

To illus­trate the fea­ture to query con­sump­tion by envi­ron­ments, we have selected the envi­ron­ment Con­tainer 1, which houses three air con­di­tion­ing units.

Table 3 presents the data obtained by the remote man­age­ment sys­tem about the energy con­sumed in this envi­ron­ment dur­ing the period of this sam­pling.

Table 3: Data Con­sump­tion of Con­tainer 1

Fig­ure 9 presents a line graph that shows the vari­a­tion of power con­sump­tion in this envi­ron­ment (Con­tainer 1) on week­ends (begin and end of the chart) and week­days (mid part).

Fig­ure 9: Graph of Con­tainer 1 Elec­tric­ity Con­sump­tion

The remote sys­tem also gen­er­ated the graph of the energy con­sump­tion of the three air con­di­tion­ers installed in this envi­ron­ment, as showed in Fig­ure 10.

Fig­ure 10: Com­par­a­tive of Elec­tric­ity Con­sump­tion in Con­tainer 1

The analy­sis of energy con­sump­tion in this envi­ron­ment showed that although there are three air con­di­tion­ers, one of them was lit­tle used com­pared to the other two. This fact sug­gests that this unit may be under­used in this envi­ron­ment. This infor­ma­tion may sug­gest that the admin­is­tra­tor inves­ti­gate the causes of this occur­rence and even decide if this equip­ment can be removed from this envi­ron­ment.

Con­sump­tion data of any envi­ron­ment is avail­able in graph­i­cal for­mat.

5.3 Simulations

Sim­u­la­tions show the expected elec­tric­ity con­sump­tion sav­ings if the air con­di­tion­ing units were turned on only dur­ing work­ing hours, for Sal­vador City Hall, it means Mon­day to Fri­day, 8 am to 5 pm.

The sim­u­la­tions devel­oped using RStu­dio project. R is an open-source soft­ware envi­ron­ment for sta­tis­ti­cal com­put­ing and graph­ics based on regres­sion analy­ses. RStu­dio project ren­ders the use of R eas­ier and more pro­duc­tive [Verzani 2011].

The remote man­age­ment plat­form allows to export col­lected data into .csv for­mat, which can be con­sumed by many ana­lyt­i­cal soft­ware. In order to ana­lyze the behav­ior of the air con­di­tion­ers, a week’s teleme­tries were gath­ered, exported and treated in Excel, to cre­ate a dataset. That set of val­ues were imported to a R pro­gram­ming envi­ron­ment - RStu­dio, where sub­sets for each air con­di­tioner unit were gen­er­ated.

Using R func­tions, devel­oped for this analy­sis, it was pos­si­ble to obtain devi­a­tions, con­sid­er­ing the mis­use of the devices out of work­ing hours–8:00 am to 5:00 pm. The func­tion was para­me­ter­ized to detect oper­a­tion of the unit by using a thresh­old of 50 mA for the cur­rent mea­sure­ment; time was mea­sured as elapsed time in min­utes. Devi­a­tions were cal­cu­lated in kWh. Table 4 shows the con­sump­tions and the devi­a­tions per air con­di­tioner from RStu­dio.

Table 4: Devi­a­tions per Air Con­di­tion­ing Unit

The data in the Table 4 indi­cate that amongst 16 equip­ment, 11 were oper­at­ing out­side work­ing hours deter­mined by the Sal­vador’s Munic­i­pal­ity. In one case, the devi­a­tion reached 35.81%. With this infor­ma­tion, the man­ager can inves­ti­gate the causes and know the moti­va­tion, employ­ing cor­rec­tive mea­sures, if applic­a­ble.

The graph in Fig­ure 11 shows the expected con­sump­tion after pro­gram­ming the hourly pro­files in rela­tion to the actual con­sump­tion mea­sured.

Fig­ure 11: Devi­a­tions Evi­denced

Results show 107.290 kWh of total devi­a­tion that cor­re­sponds to 14.14% of the total (758.621 kWh) con­sump­tion in the test period. The man­age­ment plat­form allows users to pro­gram hourly pro­files that will gen­er­ate a cost reduc­tion cor­re­spond­ing to the devi­a­tion.

Elec­tric­ity con­sump­tion sav­ings at this unit was approx­i­mately US$ 127.26, con­sid­er­ing the June-Sep­tem­ber 2019 test period.

5.4 Observed Behaviors

The ana­lyzes obtained in this study pointed out some inap­pro­pri­ate user behav­iors that should be avoided. We used Root Cause Analy­sis to address these behav­iors.

Accord­ing to Rooney and Van­den Hau­vel [2004], root causes are spe­cific under­ly­ing causes that can iden­tify which behav­iors man­age­ment has con­trol to fix and pro­pose rec­om­men­da­tions for pre­vent­ing recur­rences.

In this sense, a group com­posed of ana­lysts, admin­is­tra­tors, and advi­sors was formed to dis­cuss the causes inap­pro­pri­ate behav­iors. The group con­ducted a root cause analy­sis using cause-effect dia­gram on the var­i­ous behav­iors and has made some rec­om­men­da­tions to mit­i­gate these devi­a­tions.

Table 5 lists the cause-effect dia­gram of the prin­ci­pal occur­rences of inap­pro­pri­ate behav­iors, the root causes of these events, and their con­se­quences.

From the facts found and the main causes iden­ti­fied, the group elab­o­rated the fol­low­ing behav­ioral rec­om­men­da­tions:

1. Occur­rences 1, 3, and 4 can be mit­i­gated through the imple­men­ta­tion of the sys­tem under study, as it pro­vides for a process for automat­ing the sched­uled oper­a­tion of air con­di­tion­ing units in munic­i­pal pub­lic admin­is­tra­tion facil­i­ties;

2. Occur­rences 2 and 5 are related to users’ aware­ness of the impor­tance of man­ag­ing energy con­sump­tion in the insti­tu­tion. There­fore, the sup­port of man­agers and admin­is­tra­tors of the facil­i­ties becomes of fun­da­men­tal impor­tance for incor­po­rat­ing these processes in the orga­ni­za­tional cul­ture.

Table 5: Cause-effect Dia­gram

Two fac­tors are crit­i­cal for the suc­cess­ful imple­men­ta­tion of this type of sys­tem: strong upper man­age­ment sup­port and ade­quate sys­tem man­age­ment.

6. Discussion and Conclusions

This study was use­ful to ver­ify energy con­sump­tion pro­files, assess poten­tial waste causes, and plan the opti­miza­tion of the air con­di­tion­ing unit’s oper­a­tion for the low­est elec­tric­ity con­sump­tion.

This sec­tion high­lights the ben­e­fits, and chal­lenges faced when imple­ment­ing this sys­tem. Addi­tion­ally, we pro­vide a sys­tem’s gen­eral eval­u­a­tion and finally, syn­the­size our con­clu­sions.

6.1 Benefits

The ben­e­fits of imple­ment­ing this IoT solu­tion related to elec­tric­ity sav­ings involve the qual­ity of elec­tric­ity sup­ply; the dis­cov­ery of the peak con­sump­tion peri­ods; and under­stand­ing the behav­ior of users regard­ing the use of this resource.

The mon­i­tor­ing of these sig­nals sub­si­dized the ana­lyzes required for informed deci­sion mak­ing of inter­ven­tions to improve the sys­tem’s sta­bil­ity and pro­mote the reduc­tion of con­sump­tion.

Also, the sys­tem has demon­strated its abil­ity to pro­tect air con­di­tion­ers, mon­i­tor the qual­ity of the power sup­plied, proac­tively con­trol con­sump­tion, and estab­lish appro­pri­ate user behav­iors for reduc­ing con­sump­tion.

The elab­o­ra­tion of strate­gies to reduce elec­tric­ity con­sump­tion in this gov­ern­ment insti­tu­tion, through the analy­sis of its use, is a mea­sure related to the sus­tain­abil­ity and econ­omy of pub­lic resources.

6.2 Challenges

This study iden­ti­fied that the chal­lenges of the imple­men­ta­tion of a mon­i­tor­ing and con­trol sys­tem using sen­sor mod­ules for read­ing energy con­sump­tion of air con­di­tion­ing units lie on the human behav­ior domain since tech­ni­cally, the solu­tion is sim­ple and straight­for­ward. Observ­ing users’ behav­iors and the con­sump­tion sta­tis­tics obtained through the mea­sure­ments, we can relate the fol­low­ing behav­ioral dif­fi­cul­ties:

1. Par­a­digm shift—User loses the con­trol of turn­ing the air con­di­tioner on and off;

2. Remote man­age­ment soft­ware oper­a­tion required—Hours of oper­a­tion must be pro­grammed in the unit by a user, which needs to be reg­is­tered in the sys­tem and given “edi­tor” per­mis­sion.

3. The dif­fi­culty of oper­at­ing the remote man­age­ment soft­ware—Not all Sal­vador City ser­vants can oper­ate com­puter sys­tems, how­ever, easy their oper­a­tion. In these cases, there may be resis­tance to con­trols pre­vi­ously pro­grammed by soft­ware.

4. Lack of aware­ness of the need to improve the uti­liza­tion of pub­lic resources with elec­tric­ity costs—Some pub­lic ser­vants may not under­stand the need for the pub­lic insti­tu­tion to reduce energy costs. Need to raise employ­ees’ aware­ness of energy sav­ings and bet­ter use of nat­ural resources;

5. Unnec­es­sary oper­a­tion of air con­di­tion­ing units in the event of unplanned events—These occur­rences may harm energy sav­ings. For exam­ple, the facil­ity does not open in a day that it is sup­posed to open, and the sys­tem man­ager fails to change the unit sched­ule. The air con­di­tion­ing sys­tem will oper­ate nor­mally, caus­ing a waste of resources.

6.3 General Evaluation of the System

The sys­tem was sta­ble and appro­pri­ately col­lected and stored sen­sor data.

The Mesh net­work proved to be very effi­cient, and the IoT tech­nol­ogy was con­sid­ered ade­quate to the over­all sys­tem oper­a­tion and engi­neer­ing. At all times, latency and degra­da­tion data traf­fic on the Inter­net net­work was not detected. Cloud servers met expec­ta­tions for data cen­ter oper­a­tions and avail­abil­ity for data stor­age and retrieval.

The remote man­age­ment soft­ware con­tains the nec­es­sary func­tion­al­i­ties for para­me­ter­i­za­tion, user per­mis­sion grant­ing, and pro­gram­ming of its main func­tions. Users deemed the oper­a­tion of the sys­tem to be easy and intu­itive.

Although the infor­ma­tion avail­able through tables and graphs is easy to under­stand and inter­pret, the queries have some lim­i­ta­tions. The sys­tem allows the export of data in other appli­ca­tions’ for­mat such as spread­sheets; how­ever, a group­ing fea­ture is not avail­able so that oper­a­tor can select net­work groups or insti­tu­tions that may need to be ana­lyzed. For exam­ple, it is not pos­si­ble to select all Sec­re­tary of Edu­ca­tion facil­i­ties and group them, so the soft­ware can per­form sim­ple sta­tis­tics such as obtain­ing of the total con­sump­tion of these facil­i­ties, their envi­ron­ments, and indi­vid­ual air con­di­tion­ing units. The workaround is to export the data for each unit to a for­mat like Excel and com­pile man­age­ment sta­tis­tics else­where. Another lim­i­ta­tion of the sys­tem is that it does not dis­play the time­stamps when queries and reports are exe­cuted. Keep­ing tem­po­ral ref­er­ences is impor­tant, for com­par­i­son, tem­po­ral analy­sis, and audit­ing. Requests for changes have already been placed with the ven­dor, and these short­com­ings may be solved in future releases, so the sys­tem may become a viable acqui­si­tion for Sal­vador City Hall.

6.4 Conclusions

This study pro­vided the Admin­is­tra­tion with impor­tant knowl­edge about elec­tric­ity con­sump­tion in Munic­i­pal facil­i­ties and offered sup­port­ing data for informed deci­sion-mak­ing before inter­ven­tion in the air con­di­tion­ing sys­tems. Proac­tive inter­ven­tions make it pos­si­ble to cor­rect fail­ures and deter­mine the best dis­tri­bu­tion of air con­di­tion­ing units across munic­i­pal admin­is­tra­tion facil­i­ties. Air con­di­tion­ing usage guide­lines and rules can be cre­ated for reduc­ing energy con­sump­tion.

With the aid of mea­sur­ing plat­forms, it is pos­si­ble to check con­sump­tion pro­files, gauge poten­tial waste, and enable cost reduc­tion.

Another impor­tant fac­tor is the amount of energy avail­able, which has a direct impact on device per­for­mance, con­tribut­ing or degrad­ing its use­ful life. The pro­posed teleme­try plat­form includes sys­tems that allow to locally assess the qual­ity of the dis­tri­bu­tion net­work, in addi­tion to eval­u­at­ing the pric­ing mea­sures.

More­over, processes con­trol within the facil­ity assumes spe­cial rel­e­vance in the con­trol of elec­tric­ity resources. The sys­tem pro­vides auto­mated mon­i­tor­ing and con­trol of energy con­sump­tion, whereas the deci­sion-mak­ing process of the facil­ity’s admin­is­tra­tor plays a rel­e­vant role in the effec­tive­ness of this process.

Con­sumer habits should also be con­sid­ered. Edu­ca­tional cam­paigns should be imple­mented to advise the space users to turn lights and air con­di­tion­ers off in the absence of peo­ple. A sim­ple mea­sure is to place stick­ers close to the light switch remind­ing peo­ple to do so.

7. Future Studies

Using the same IoT infra­struc­ture cre­ated for this study and sen­sors avail­able in the mar­ket, the Admin­is­tra­tion envi­sions the instal­la­tion of other types of sen­sors to per­form new mea­sure­ments such as the gen­eral con­sump­tion of elec­tric­ity and water.

Elec­tric­ity meters installed on switch­boards could be cer­ti­fied by the con­ces­sion­aire to stream­line the billing process. Elim­i­nat­ing man­ual read­ings of local meters and elec­tron­i­cally exchang­ing infor­ma­tion may reduce costs for con­sumers and the util­ity. This fea­ture can be imple­mented elec­tron­i­cally using a Gov­ern­ment to Busi­ness (G2B) e-gov­ern­ment appli­ca­tion.

The analy­sis of data col­lected from auto­mated water con­sump­tion meters may pro­vide sup­port for actions to mit­i­gate wast­ing this resource. The detec­tion of abrupt con­sump­tion vari­ance, over any period, can lead to imme­di­ate cor­rec­tive actions such as inter­rup­tion of water sup­ply through auto­mated valves, or sig­nal­ing for repairs to the hydraulic net­work. The elab­o­ra­tion of con­sump­tion pro­files of pub­lic build­ings and places, espe­cially those that rely on pub­lic water sources is yet another prod­uct of this type of project.

Acknowledgments

Fed­eral funds sup­ported this work through FCT—Fun­dação para a Ciên­cia e Tec­nolo­gia within the Pro­ject Scope: UID/CEC/00319/2019. This work also had the col­lab­o­ra­tion of the com­pany MOD­U­LUS ONE and the Sec­re­tariat of Man­age­ment of Sal­vador City Hall (SEMGE).

References

• ANEEL (2018). Pro­ced­i­men­tos de Dis­tribuição de Ener­gia Elétrica no Sis­tema Elétrico Nacional—PRODIST. Agên­cia Nacional de Ener­gia Elétrica, Brasil. Retrieved Sep­tem­ber 23, 2019 from http://www.aneel.gov.br/doc­u­ments/656827/14866914/Módulo_8-Revisão_10/2f7cb862-e9d7-3295-729a-b619ac6baab9

  AWS (2019). Ama­zon Com­pute Ser­vice Level Agree­ment. Ama­zon Web Ser­vices. Retrieved Sep­tem­ber 24, 2019 from https://aws.ama­zon.com/com­pute/sla/

  Almarabeh, T., Y. Kh. Maj­dalawi and H. Moham­mad (2016). Cloud Com­put­ing of E-Gov­ern­ment. Com­mu­ni­ca­tions and Net­work 8(1), 1–8. DOI:https://doi.org/10.4236/cn.2016.81001

  Bhati, A., M. Hansen and C. M. Chan (2017). Energy con­ser­va­tion through smart homes in a smart city: A les­son for Sin­ga­pore house­holds. Energy Pol­icy 104, 230–239. DOI: https://doi.org/10.1016/j.enpol.2017.01.032

  Chodorow, K. (2019). Mon­goDB: The Defin­i­tive Guide, sec­ond edi­tion. Sebastopol, CA: O’Reilly.

  Gil-Gar­cia, J. R., N. Hel­bigb and O. Adeg­boyega (2014). Being smart: Emerg­ing tech­nolo­gies and inno­va­tion in the pub­lic sec­tor. Gov­ern­ment Infor­ma­tion Quar­terly 31, I1–I8. DOI:https://doi.org/10.1016/J.GIQ.2014.09.001

  Kha­je­nasiri, I., A. Esteb­sari, M. Ver­helst and G. Gie­len (2017). A Review on Inter­net of Things Solu­tions for Intel­li­gent Energy Con­trol in Build­ings for Smart City Appli­ca­tions. Energy Pro­ce­dia 111, 770–779. DOI:https://doi.org/10.1016/j.egypro.2017.03.239

  Klein­baum, D. G. and L. L. Kup­per (1978). Applied Regres­sion Analy­sis and Other Mul­ti­vari­ate Meth­ods. North Sci­t­u­ate, Mass­a­chu­setts: Duxbury Press.

  Kyr­i­azis, S., T. Var­varigou, A. Rossi, D. White and J. Cooper (2013). Sus­tain­able smart city IoT appli­ca­tions: Heat and elec­tric­ity man­age­ment & Eco-con­scious cruise con­trol for pub­lic trans­porta­tion. 2013 IEEE 14^th Inter­na­tional Sym­po­sium on “A World of Wire­less, Mobile and Mul­ti­me­dia Net­works” (WoW­MoM), 1–5. DOI: https://doi.org/10.1109/WoW­MoM.2013.6583500

  Omura, H. (2000). Infor­ma­tion Tech­nol­ogy (IT) for E-Gov­ern­ment. FUJITSU Sci­en­tific & Tech­ni­cal Jour­nal 36(2), 232–235.

  Park, S. H., S. Cho and J. R. Lee (2014). Energy-effi­cient prob­a­bilis­tic rout­ing algo­rithm for inter­net of things. Jour­nal of Applied Math­e­mat­ics 2014(1). DOI: https://doi.org/10.1155/2014/213106

  Poko­rny, J. (2013). NoSQL data­bases: A step to data­base scal­a­bil­ity in web envi­ron­ment. nter­na­tional Jour­nal of Web Infor­ma­tion Sys­tems
  9(1), 69–82. DOI: https://doi.org/10.1108/17440081311316398

  Ramya, R. and K. Ramya (2015). Cloud com­put­ing. Inter­na­tional Jour­nal on Appli­ca­tions in Elec­tri­cal and Elec­tron­ics Engi­neer­ing 1(8), 1–5.

  Rooney, J. J. and L. N. Van­den Hau­vel (2004). Root Cause Analy­sis For Begin­ners. Amer­i­can Soci­ety for Qual­ity 7, 45–53. Retrieved Sep­tem­ber 2, 2019 from http://asq.org/qual­ity-progress/2004/07/qual­ity-tools/root-cause-analy­sis-for-begin­ners.pdf

  Schu­ur­man, D. and P. Tõnurist (2017). Inno­va­tion in the Pub­lic Sec­tor: Explor­ing the Char­ac­ter­is­tics and Poten­tial of Liv­ing Labs and Inno­va­tion Labs. Tech­nol­ogy Inno­va­tion Man­age­ment Review 7(10, 7–14. DOI:https://doi.org/10.22215/tim­re­view/1045

  Seifert, J. W. (2003). A Primer on E-Gov­ern­ment: Sec­tors, Stages, Oppor­tu­ni­ties. CRS Report for Con­gress. The Library of Con­gress.

  Vas­sileva, I. and J. Campillo (2016). Con­sumers’ per­spec­tive on full-scale adop­tion of smart meters: A case study in Västerås, Swe­den. Resources 5 (1). DOI: https://doi.org/10.3390/resources5010003

  Verzani, J. (2011). Get­ting Started with RStu­dio. Sebastopol, CA: O’Reilly.

  Wang, X. and A. O. Lim (2008). IEEE 802.11s wire­less mesh net­works: Frame­work and chal­lenges. Ad Hoc Net­works 6(6), 970–984. DOI: https://doi.org/10.1016/j.adhoc.2007.09.003

  Zheng, J., D. W. Gao and L. Lin (2013). Smart Meters in Smart Grid: An Overview. 2013 IEEE Green Tech­nolo­gies Con­fer­ence (Green­Tech), 57–64. DOI:https://doi.org/10.1109/Green­Tech.2013.17