Rethinking Collection System Cleaning Using Acoustic Inspection

Transcription

1 WEFTE 2012 Sesson 114 O&M of ollecton Systems Rethnkng ollecton System leanng Usng Acoustc Inspecton Ivan Howtt 1*, John Fshburne 2 1 InfoSense, Inc., harlotte, North arolna. 2 harlotte-mecklenburg Utltes, harlotte, North arolna. * Emal: howtt@nfosensenc.com ABSTRAT Settng up a utlty s preventatve mantenance program for ther gravty-fed collecton system s cleanng nvolves recognzng that many ppe segments scheduled for cleanng wll not need t. But, ths cost of over cleanng must be balanced aganst nsuffcent mantenance resultng n excessve overflows. We present heren a new dagnostc tool, the Sewer Lne Rapd Assessment Tool or SL-RAT. The SL-RAT s an onste nspecton devce developed by InfoSense wth the support of harlotte-mecklenburg Utltes (MU). The SL-RAT dagnostc capablty allows cleanng requrements for ppe segments to be economcally prortzed pror to conductng cleanng operatons. We show that sgnfcant cost savngs can be acheved through: reduced asset condton assessment costs, reduced cleanng-related overflows, and reduced nonvalue added cleanng effort. In ths paper, MU ase Studes are revewed and used to show how the SL-RAT acoustc nspecton system can be used to establsh a Sewer Lne ondton Based Mantenance (SL-BM ) program. KEYWORDS Sewer lne nspecton, wastewater operaton management, ppe blockage assessment, collecton system cleanng, condton based mantenance, acoustc ppe nspecton. INTRODUTION Based on harlotte Mecklenburg Utltes (MU) ongong efforts to enhance ther gravty-fed collecton system s mantenance operatons, t has been recognzed that 50% or more of ppe segments cleaned receve lttle to no beneft as they were essentally already clean pror to mantenance. Ths nsght s further compounded by the fact that locatons chosen for conductng cleanng operatons were dentfed usng spatal analyss as well as hstorcal performance and nsttutonal system knowledge as nputs. ollecton system mantenance requres allocatng cleanng resources to the rght place pror to system falure (SSO, manlne blockages and buldng backups). MU s overall overflow rate has been between 8 and 9 overflows/100 mles/year wth over 20% of the system cleaned annually. For comparson, a scatter plot of overflows/100 mles vs. percentage system cleaned s depcted n Fgure 1. The plot s based on self reportng from sxteen muncpaltes annual performance reports. As would be antcpated, lnear regresson ndcates a strong correlaton between cleanng effort and overflow reducton. Due to the nherent random nature of the underlyng mechansms that buld up to overflows, there s lkely a dmnshng return wth more cleanng. Therefore, as the percentage of the system cleaned ncreases, an even larger proporton of wasteful and unnecessary cleanng wll be conducted, unless, an mproved method for targetng cleanng operatons can be mplemented.

2 Overflows / 100m WEFTE 2012 Sesson 114 O&M of ollecton Systems harlotte 45% Lnear Regresson Standard Devaton: % % System leaned Annually One Sgma Below Mean Fgure 1 omparson of sxteen muncpaltes performance n mantanng ther collecton system s overflow rates based on the percentage of the system cleaned annually. Hstorcally, usng condton based nspecton to determne where and when to deploy collecton system cleanng resources has not been economcally feasble. Exstng ppe nspecton methods are ether too cost prohbtve for wde spread use or provde nadequate condton assessment. The Sewer Lne Rapd Assessment Tool or SL-RAT s a new dagnostc tool specfcally desgned to address the need for dentfyng ppe blockage n support of an overall collecton system mantenance program. The SL-RAT s an onste nspecton devce developed by InfoSense wth the support of MU. The SL-RAT dagnostc capablty allows cleanng requrements for ppe segments to be economcally prortzed pror to conductng cleanng operatons. The acoustc testng equpment provdes a clear condton assessment drectly correlated wth cleanng requrements. The acoustc nspecton operaton requres sgnfcantly less resources than s requred for current mantenance practces. Ths provdes the opportunty to rethnk usng condton based mantenance as a vable tool for deployng cleanng resources. Ths novel approach can both mprove mantenance qualty and reduce unnecessary mantenance operatons. Over the past two years, MU has conducted well over three thousand SL-RAT acoustc nspectons of ppe segments. Ths has provded a condton assessment for over 850 thousand feet of ther collecton system. In the past year, the focus of nspectons has changed from evaluatng the SL-RAT technology, to ncorporatng the technology as the enabler for a newly evolvng collecton system cleanng BM program based on acoustc nspectons. The objectve of the paper s to provde: 1. Drect evaluaton of the SL-RAT as a condton based assessment tool. Ths evaluaton s based on a plot project conducted by MU to evaluate the SL-RAT s effectveness n prortzng mantenance operatons, as well as to evaluate ts cost effectveness and ease of operaton. Ths study nvolved a systemc comparson between acoustc nspecton and correspondng TV assessment both before and after mantenance operatons. 2. Drect evaluaton of ntegratng the SL-RAT for condton based assessment. An llustratve case study s presented to examne the procedural and operatonal advantage

3 WEFTE 2012 Sesson 114 O&M of ollecton Systems for usng the SL-RAT as a condtonal based assessment tool for a consoldated cleanng project. 3. Evaluate the capablty and requrements for developng a Sewer Lne ondton Based Mantenance (SL-BM ) program based on SL-RAT acoustc nspectons. Examne the trade-offs n developng a SL-BM program to acheve a collecton system performance goal of 2 overflows/100 mles/year wthout ncreasng the current operatons cost for cleanng. 4. Evaluate the ost versus Performance for an SL-BM usng the SL-RAT based on the spatal and temporal overflow patterns wthn the MU collecton system. Ths evaluaton s based on an analytcal model whch ncorporates the MU overflow data for the past thrteen years and the SL-RAT acoustc nspecton data collected over the past two years. METHODOLOGY Acoustc Inspecton SL-RAT Operaton The sewer-lne rapd assessment tool (SL-RAT ) explots the smlartes and dfferences between water and sound transmsson through a sewer lne segment n order to dagnose the extent of the ppe s blockage. Ths novel, patented methodology (Howtt 2009) s based on measurng the sgnal receved from an actve acoustc transmsson through a segment. Fgure 2 depcts the general confguraton of the SL-RAT devce. The acoustc transmtter generates sound waves just below the entrance to the manhole whch naturally couple nto the connectng sewer lne segments, whether the depth of the manhole s 3 feet or greater than 20 feet. The sound wave propagates n the ar gap above the wastewater flow from the speaker to the recevng mcrophone located at the adjacent manhole. Segment lengths exceedng 700 feet have been successfully evaluated. The acoustc recever measures the acoustc plane wave from the transmtted sgnal n order to evaluate the condton of an entre segment and provdes an onste assessment n less than three mnutes. An mportant practcal aspect of the SL-RAT s that both the speaker and the mcrophone are placed just wthn the openng of the manhole and never come n contact wth the wastewater flow and the operators have no requrement for confned space entry. Acoustc Transmtter Acoustc Recever 6 Dameter or Greater Sewer Lne Segment Fgure 2 oncept and operaton of the SL-RAT Acoustc Inspecton System.

4 WEFTE 2012 Sesson 114 O&M of ollecton Systems A ppe segment s a natural acoustc wavegude (Phlp 1968). As llustrated n Fgure 2, commonly encountered santary sewer defects, such as roots, grease, ppe sags and ppe breakages naturally absorb or reflect acoustc energy. These defects change a segment s acoustc propertes and produce a measurable mpact on the receved sgnal at the mcrophone,.e., the segment s acoustc fngerprnt (SAF). Each segment has an ndvdual SAF representatve of ts current state. The SAF changes over tme as the condton of the segment vares. The SL-RAT uses the SAF to determne the SL-RAT Blockage Assessment,.e., an estmate of the aggregate blockage wthn the ppe segment between the acoustc transmtter and acoustc recever. The aggregate blockage assessment s provded to the operator at the end of each test on a scale from 0-10 wth zero ndcatng complete blockage and ten ndcatng an essentally clean segment. Usng the blockage assessment, the operator can determne whether or not mantenance s needed. ase Studes MU Plot Project ntegrated the SL-RAT evaluaton nto an ongong preventatve mantenance program whch focuses cleanng and nspecton resources nto a consoldated area. A detaled presentaton of both the approach and the results of the plot project are presented n Howtt and Fsburne (2010). Key ponts from ths paper are summarzed here due to ther mportance n valdatng the SL-RAT and to establsh ts operatonal performance. The performance results from ths study are mportant for developng and motvatng the SL-BM program based on the SL-RAT. MU Plot Project objectves: 1. Assess the ablty of the SL-RAT acoustcal tool to a. Reveal a state of partal obstructon n a ppe segment, b. Defne the progresson of partal to complete obstructon for a ppe segment. 2. Establsh the sutablty of the SL-RAT acoustcal tool for MU to realze operatng cost savngs and prevent system falures a. Economcally dentfy ppe segments requrng mantenance (prortzaton ad, falure preventon method), b. Establsh optmal mantenance cycles (sustanable falure preventon). Approach used n evaluatng the MU Plot Project objectves: 1. Use the followng data collecton sequence wthn the onsoldated leanng Area: a. ollect pre-cleanng acoustcal profle of ppe segments b. ollect pre-cleanng vdeo nspecton of ppe segments c. ollect post-cleanng vdeo nspecton of ppe segments d. ollect post-cleanng acoustcal profle of ppe segments 2. Analyze and assess acoustcal profles and compare wth assessment based on vdeo nspectons. 3. Evaluate SL-RAT use by MU feld staff to dentfy reduced capacty wthn ppe segments wthout the beneft of havng a prevous acoustcal profle a. Defne and analyze range of profles for clean ppes, b. SL-RAT classfcaton analyss: False Postve Rate and False Negatve Rate c. ost savng analyss.

5 WEFTE 2012 Sesson 114 O&M of ollecton Systems The ppe segments to be evaluated by the SL-RAT acoustc nspecton were based on the precleanng requrements for the basn cleanng project. MU has found that pre-cleanng needs to be an ntegral part of a consoldated cleanng project n order to reduce the rsk of splls durng ther systematc cleanng operaton. Pror to Basn cleanng operatons, an Engneerng Assessment was made on whch ppe segments were at hgh rsk of beng obstructed. These ppe segments were then used to defne the ppe segments ncluded n the SL-RAT acoustc testng as ndcated n Fgure 3 by the hghlghted ppe segments. The pre-cleanng classfcaton results n Fgure 3 are based on usng the SL-RAT Blockage Assessment to evaluate the pre-cleanng classfcaton. The SL-RAT acoustc nspecton successfully dentfed the ppe segments requrng pre-cleanng. The number of ppe segments dentfed was approxmately one-thrd that dentfed orgnally usng MU s standard method of Engneerng Assessment usng hstorcal data. These results were confrmed usng TV, as presented n the followng dscusson. Fgure 3 GIS map for Basn , MU Plot Project locaton. Hghlghted ppe segments represent locatons evaluated wth the SL-RAT acoustc nspecton. The three TV snapshots llustrate comparson between the SL-RAT Blockage Assessment and the TV Blockage Assessment. An mportant aspect of the methodology employed n the MU Plot Project was collectng TV vdeos both pror-to and after cleanng the ppe segments. Ths approach allowed the techncal feasblty of the SL-RAT to be valdated. In order to make a comparatve analyss between the TV and the SL-RAT, the TV vdeos were ndependently revewed to assess the aggregate blockage wthn each ppe segment,.e., TV Blockage Assessment. A rank order was used n the assessment wth 10 ndcatng a ppe segment was essentally clean and 0

6 WEFTE 2012 Sesson 114 O&M of ollecton Systems ndcatng a ppe segment was substantally obstructed. Three examples are gven n Fgure 3 comparng the TV Blockage Assessment wth the SL-RAT Blockage Assessment wth an accompanyng TV snapshot to llustrate the condton of the ppe segment. The scatter plot n Fgure 4 provdes a comparson based on the two assessments. Each pont on the plot represents the TV Blockage Assessment versus the SL-RAT Blockage Assessment for one ppe segment. If the two assessment where dentcal, then all the ponts would le along the dagonal lne from (0, 0) to (10, 10). As observed n Fgure 4, almost all ponts n the scatter plot le n the trangle below the dagonal. Ths mples the SL-RAT Blockage Assessment wll tend to be more conservatve than the TV Blockage Assessment. The scatter plot also ndcates a threshold can be establshed for the SL-RAT Blockage Assessment whch allows us to dscrmnate between essentally clean and essentally drty ppe segments. Ths s ndcated as the SL-RAT Standard Threshold n Fgure 4. Usng ths threshold, all ppe segments wth a SL-RAT Blockage Assessment less than the threshold would requre cleanng and all those above would be classfed as suffcently clean and not requre addtonal mantenance. Based on the ppe segments evaluated for the MU Plot Project, usng the SL-RAT Standard Threshold, a 61% reducton n the cleanng requrement s recommended. A second threshold s ndcated on the graph, SL-RAT rtcal Threshold. Ths threshold dscrmnates between ppe segments n crtcal need of mantenance and those that are not. Usng ths threshold ndcates 85% of the ppe segments n the MU Plot Project are not n need of mmedate mantenance, based on the acoustc nspecton. On the other hand, 15% should have mmedate acton taken, whether t s cleanng or a more detaled nspecton such as TV. The results recommended usng the SL- RAT Thresholds are consstent wth the TV Blockage Assessments. For both of the SL-RAT Thresholds consdered, there s no falure n dentfyng a ppe segment whch requred further mantenance as one that dd not. To estmate the SL-RAT cost savngs, we need to estmate the cost for operatng the SL-RAT by a two person feld crew. The assumptons used are gven n Table 1 where the values n the table reflect typcal ndustry values. The onste work hours per day has been selected conservatvely. The average number of segments nspected per hour usng the SL-RAT s based on MU/InfoSense feld experence usng the SL-RAT durng the course of the MU Plot Project. The nspected segments per hour vared from 4 to 10 wth 6 beng the average. Ths productvty estmate contnues to hold based on MU s extensve use of the SL-RAT wth several dfferent feld crews. Usng the values from Table 1, the SL-RAT operatonal cost per foot s $0.09/ft (1)

7 SL-RAT Blockage Assessment WEFTE 2012 Sesson 114 O&M of ollecton Systems SL-RAT Standard Threshold 4 2 SL-RAT rtcal Threshold TV Based Blockage Assessment Fgure 4 Scatter plot comparng the TV Blockage Assessment wth the SL-RAT Blockage Assessment for sewer lne segments wthn the MU Plot Project. SL-RAT Thresholds used to dscrmnate between those ppe segments requrng addtonal mantenance (below the threshold) and those that do not (above the threshold). Table 1 Assumpton Used for Evaluatng SL-RAT ost Based on the MU Plot Project Number of rew Members 2 Annual Fully Loaded Salary $68,000 Annual Equpment osts (Includng Truck & SL-RAT) $24,000 Work Days Per Year 251 Onste Work Hours Per Day 5.5 SL- RAT Average Number of Segments Inspected Per Hour 6 Average Sewer Lne Segment Length n feet 220 ost Per Foot $0.09/ft Next, n order to evaluate the SL-RAT cost-performance, we need to frst estmate the cost of the cleanng operaton wthout the SL-RAT. The MU urrent Basn leanng Operaton s based on cleanng an entre basn followed by TV nspecton to verfy the cleanng operaton. For the evaluaton we have used one dollar as the average costs per foot ($1.00/ft) for both the TV crew nspecton and the cleanng crew operaton. To generalze the results, we have used typcal

8 WEFTE 2012 Sesson 114 O&M of ollecton Systems ndustry cleanng and TV costs and not specfc values from the MU Basn leanng Project. To smplfy the evaluaton, we have excluded the cost for easement cleanng and clearng. Usng these assumptons, the urrent Basn leanng Operaton cost per foot of sewer lne ppe s $2.00/ft. The cost savng analyss for the SL-RAT Based Basn leanng Operaton needs to balance the cost ncrease assocated wth usng the SL-RAT wth the cost savngs obtaned by reducng the number of segments to be cleaned. Ideally, we would elmnate all segments requrng mnmal mantenance. Usng the SL-RAT Standard Threshold reduces the number of ppe segments by 61%. In addton, usng ths SL-RAT threshold, no ppe segments requrng cleanng are mssed. The cost savng s drven by the dfference between the total cost estmated for the MU urrent Basn leanng Operaton and the total cost estmated for the SL-RAT Based Basn leanng Operaton: MUurrent Basn leanng SL - RAT Based Basn leanng ost Savngs $2.00/ft $0.09/ft $1.13/ft Number Feet. $2.00/ft Number Feet, (2) $0.09/ft $2.00/ft Number Ft., 0.39 $2.00/ft Number Feet (3) (4) Basn leanng Project was conducted to further test and to fully mplement the use of the SL- RAT wthn a basn cleanng project. The goal of the project was to use acoustc nspecton to dscrmnate between ppe segments requrng cleanng and those that do not (harlotte 2011). A 52,000 ft basn wthn MU s collecton system was selected. The basn selected was scheduled for consoldated cleanng whch would typcally requre the entre basn to be cleaned and then vdeoed wth TV. Instead, acoustc nspecton was used to assess the cleanng requrements wthn the entre basn pror to cleanng. A more conservatve SL-RAT Standard Threshold was used for the project,.e., the threshold was set to 6 rather than 4 as used n the MU Plot Project. The results of the acoustc nspecton are depcted n Fgure 5. Wth the acoustc nspecton results, cleanng was reduced to 22,000 ft (42% of the basn). The 30,000 ft (58%) of sewer whch passed the condtonal assessment requred no mantenance,.e., cleanng or post cleanng verfcaton. In addton, seven of the eleven segments classfed as "Blocked" were found to be full ppe sags and were scheduled for repar. Usng the same cost analyss as gven n equatons (2) (4), the cost savngs for the Basn leanng Project was ost Savngs $2.00/ft $0.09/ft $1.07/ft 52,000 $55, $2.00/ft Number Feet (5)

9 WEFTE 2012 Sesson 114 O&M of ollecton Systems Fgure 5 SL-RAT acoustc nspecton assessment for Basn leanng Project. Pror to cleanng 58% of the basn classfed as lean and no further mantenance acton was requred. ONDITION BASED MAINTENANE PROGRAM Overvew Mantenance polces for wastewater collecton systems cleanng operatons are currently a combnaton of fxed nterval mantenance,.e., Tme-Based Mantenance (TBM) and reactve mantenance,.e., orrectve Mantenance (M). Fgure 6(a) llustrates the optmal regon of applcaton for each strategy. The horzontal axes represents the remanng tme to falure wth values decreasng towards the rght. The vertcal axs represents the relatve rsk and the cost assocated wth a ppe segment overflow. To llustrate, vandalsm can lead to overflows, e.g., dumpng leaves n a manhole. Snce vandalsm s an unlkely event and the tme to falure s short, a M program s the only opton. A TBM program s approprate n areas where perodc cleanng nterval s requred and can be relably estmated, e.g., areas wth hgh grease restaurants. In these areas there s a hgh rsk and the tme nterval to falure can be predcted. A preponderance of grease and root blockages occur over a suffcently long tme nterval, suggestng a BM program s optmal. From Wseman et. al., ondton based montorng s defned as: an dentfable physcal condton whch ndcates that a functonal falure s ether about to occur or n the process of occurrng. In ths process, the tems are nspected and left n servce on the condton that they meet specfed performance standards. The frequency of these

10 Overflow Rsk & ost SL-Segment ondton WEFTE 2012 Sesson 114 O&M of ollecton Systems nspectons s determned by the potental falure (P-F) nterval, whch s the nterval between the emergence of the potental falure and ts decay n to a functonal falure. Developng an overall mantenance polcy that balances the mantenance strateges s the goal of Relablty entered Mantenance (RM) (Moubray 1997). RM allocates cleanng resources based on optmzng the cost and rsk assocated wth overflows. Max TBM P, Potental Falure D, Detect Potental Falure ΔT D M RM BM BM Inspectons F-D ΔT F-P F, Functonal Falure Mn Max Remanng Tme to Falure, ΔT (a) Mn Operaton Tme (b) Fgure 6 (a) Regons of optmal applcaton for four mantenance strateges (Lehtonen 2006); (b) Relatonshp between nspecton and P-F n a BM based mantenance strategy (Moubray 1997 and Wseman et. al.). Hstorcally, usng condton based nspecton to determne where and when to deploy collecton system cleanng resources has not been economcally feasble. The avalable nspecton technologes are ether cost prohbtve or provde nadequate nformaton. The SL-RAT acoustc nspecton equpment provdes a clear condton assessment drectly correlated wth the cleanng requrements. Acoustc nspecton requres sgnfcantly fewer resources compared to normal mantenance,.e., SL-RAT acoustc nspecton has been shown to be less than 1/10 th the cost of cleanng or just $0.09/ft. Ths provdes the opportunty to rethnk usng condton based mantenance as a vable tool for deployng cleanng resources. Ths approach can mprove mantenance qualty, reduce unnecessary mantenance operatons and, at the same tme, reduce costs. The prevous dscusson motvates the value proposton for mplementng the Sewer Lne ondton Based Mantenance (SL-BM) program based on the SL-RAT nspecton. Fgure 6(b) llustrates the concept and the challenges wth mplementaton. The graph n the fgure s a standard P-F curve (Moubray 1997 and Wseman et. al.) depctng a graceful degradaton n a ppe segment wth the condton assessment graph representatve of a grease or root mode of falure. Pont P represents the ntal tme performance degradaton can be detected and Pont D represents the tme performance degradaton s detected based on the SL-RAT BM nspecton schedule. Pont F represents the operaton tme at whch the sewer lne ppe segment functonally fals, e.g., the blockage s suffcent to cause an overflow. Each ppe segment has a

11 WEFTE 2012 Sesson 114 O&M of ollecton Systems unque P-F curve governed by underlyng factors nfluencng ts falure rate. The goal of the SL- BM s to estmate the BM nspecton and mantenance tmes to ensure mantenance s scheduled and conducted pror to the ppe segments falure at a sgnfcant cost savngs over current mantenance programs. The SL-RAT s an essental tool n developng an effectve SL-BM cleanng program. To explore the sgnfcance of an SL-BM cleanng program, we extrapolate the results from the MU Plot Project. The graph n Fgure 7 provdes a hypothetcal comparson of the cost effectveness between three cleanng programs. The purpose of the comparson s to llustrate the flexblty and trade-offs avalable n desgnng a cleanng program based on acoustc nspecton. We draw upon the results of the lnear regresson n Fgure 1 to set a performance goal for the collecton system cleanng program. From the lnear regresson, to acheve an operatons performance goal of 2 overflows/100 mles/ year, requres 76% of the collecton system to be mantaned annually. The correspondng one standard devaton below the mean requres 45% of the system to be mantaned. Ths result s translated to Fgure 8 where the blue regon hghlghts the range of possble solutons for the collecton system cleanng cost/100 mles of ppe to be wthn the correspondng 45% to 76% system mantenance requrements. In Fgure 7 at Pont-1, for the baselne cleanng polcy, 20% of the ppes are cleaned annually at $1/ft. For ths cost analyss we only consder the cost of cleanng and do not nclude any cost for post cleanng nspecton wth TV. The other two polces are based on usng the acoustc nspecton tool pror to cleanng. For these two polces, 20% of the ppes have been nspected and, based on ther SL-RAT Blockage Assessment, only 7.8% are estmated to requre cleanng resultng n over a 50% cost savngs. Only segments whch are dagnosed as essentally clean are removed from the cleanng operatons resultng n no mpact on collecton system performance. Next we look at keepng the budget fxed at the 20% annual baselne cleanng cost and look at two dfferent cleanng and nspecton polces usng the SL-RAT. For SL-BM I, we contnue usng the polcy that only essentally clean lne segments are removed from the cleanng operatons based on ther acoustc blockage assessment. Ths allows us to acoustc nspect 41.6% of the collecton system and based on the SL-RAT Blockage Assessment only 16.2% requre cleanng, Pont-2. SL-BM II takes a dfferent approach by swtchng modes to focus on only cleanng the ppe segments whch are n mmedate need,.e., only clean f dagnosed wth a sgnfcantly low SL-RAT Blockage Assessment. The polcy transfers more resources towards nspecton rather than cleanng, allowng 63.4% of the collecton system to be acoustc nspected, wth an estmated 14.1% cleaned and wth an estmated 9.5% dagnosed as havng a sgnfcant blockage assessment by the SL-RAT acoustc nspecton tool, Pont-3. Ths suggests that by usng acoustc nspecton, over 60% of the collecton system can be mantaned annually at a comparable cost as a 20% annual baselne cleanng program. Ths acheves the goal of mantanng the collecton system between 45% and 76% wthout ncreasng the annual cost. The SL-RAT acoustc condton assessment s used to cost effectvely target cleanng resources to locatons wth a hgher lkelhood to cause overflows. The prevous dscusson provdes a general assessment of SL-BM program usng the SL-RAT acoustc nspecton. Varatons n mplementaton are examned under the assumpton that overflows are equally lkely wthn the collecton system. We next turn to evaluatng the SL-

12 leanng ost / 100 m Ppe leanng ost / 100 m Ppe (Mllons $) WEFTE 2012 Sesson 114 O&M of ollecton Systems BM cost versus performance mpact takng nto account the hstorcal spatal overflow patterns wthn MU s collecton system. In addton, characterstcs of the SL-RAT performance are ncorporated based on MU s operatonal acoustc nspectons over the past two years. % ollect System Mantaned for 2 Overflows/100m Based on One Sgma Lower Bound $300, $250, $200, $150, $100, Baselne leanng Program 1 2 x 2.1 SL-BM leanng Program II SL-BM leanng Program I Total Ppe Segments n leanng Operaton (%) 3 Fgure 7 ollecton System cleanng polcy comparson between Baselne leanng Program wth two dfferent ondton Based Mantenance programs (SL-BM Program I & SL-BM Program II) whch use SL-RAT acoustc nspecton to prortze cleanng operatons. MU SSO Evaluaton Understandng the spatal and temporal dstrbuton of overflows wthn the collecton system plays an mportant role n developng and managng a mantenance program. The followng evaluaton s based on MU data collected for FY 2000 through FY 2012 (FY 2012 data s through May 24, 2012 and the MU fscal year ends June 30 th ). Fgure 8(a) shows a spatal dstrbuton of the 4,386 overflows wthn the collecton system over the past thrteen year perod. The dstrbuton s based on evaluatng the number of overflows occurrng wthn onesquare-mle based on a one-square-mle grd overlad on Mecklenburg ounty. The ounty boundares are ndcated by the dashed black lnes n the fgure. The overflows occur wthn 390 square mles of the 526 square mles of the MU collecton system. Overflow occurrence s well correlated wth collecton system ppe densty and populaton densty. Varaton n overflow densty s strongly evdent n Fgure 8(a). The overflow dstrbuton s further nvestgated by evaluatng the hstogram of the overflow occurrence wthn each onesquare-mle. The hstogram s gven n Fgure 8(b). The overflow dstrbuton has been grouped

13 Locaton (mles) Total Overflows/Square Mle Relatve Occurance WEFTE 2012 Sesson 114 O&M of ollecton Systems nto four regons as ndcated by four colors. Regon 0 ncludes the 136 one-square-mle locatons wth no reported overflows n the past thrteen years. Regon I through Regon III are defned based on the overflow ranges specfed n the fgure s legend. Above each correspondng Regon n the hstogram, the followng two tems are gven: the total number of overflows wthn the regon and the number of square mles encompassed by the regon. The regons dentfed n Fgure 8(b) help dentfy areas wthn the collecton system requrng dfferent levels of prorty for allocatng mantenance resources based on the hstorcal overflow data. To llustrate, the 136 square mles wth zero reported overflows, Regon 0, requres mnmal change n resources allocated to t or perhaps even a reducton n resources s warranted. Whereas, Regon III ncludes 80 one-square-mle locatons wth each locaton averagng over two overflows per year; Regon III locatons would be good targets for reducng the annual overflow rate. Ths segmentaton wll be further exploted n evaluatng the overflow characterstc wthn MU and n developng the MU SL-BM program based on the SL-RAT sq. m overflows /136 m 2 0 Overflows 0 < Overflows < Overflows 20 Overflows > Locaton (mles) overflows /233 m / 77 m / 80 m Spll Event Dstrbuton wthn 1 sq. m. grd (a) (b) Fgure 8 MU SSO Evaluaton for FY 00 through FY12 (a) Spatal representaton of the overflows that occurred wthn one-mle-square grd over the thrteen year perod; (b) Hstogram of the number of overflows that occurred wthn a one-mle-square over the thrteen year perod wth the bar color dstngushng between three dsjont regons. Fgure 8 provdes an aggregate hstorcal perspectve on the dstrbuton of overflow events wthn the collecton systems, but does not provde an ndcaton of when they occurred. To obtan a dfferent perspectve, the temporal occurrence of the overflows s examned. In Fgure 9, movng averages are provded based on the MU s recorded date for each overflow occurrence wthn FY 2000 to FY The movng averages are normalzed to provde the overflow rate per 100 mles of lnear feet of ppe occurrng n a year,.e., overflows/100m./year. Both three month and twelve month movng averages are depcted n Fgure 9. Seasonal varatons n the overflow rate are evdent n the three month movng average. In addton to the movng average, the past four year trend n the overflow rate s evaluated by lnear regresson over the 12 month movng average. Both the graph of the lnear regresson and the ±50% confdence for the lnear regresson s depcted n the fgure. The slope of the lnear regresson provdes a measure of the annual rate of change n the overflows/100m. The slope s -0.6,

14 Overflow Rate / 100m / Year WEFTE 2012 Sesson 114 O&M of ollecton Systems ndcatng a reducton n the overflows/100m of 0.6 for each of the past four years. The termnal date for the lnear regresson s November Evaluatng the lnear regresson for ths date, the overflow rate s 7.3 overflows/100m. Usng the same approach, temporal characterstcs of the overflows n Regons I, II and III are evaluated. Fgure 10 provdes the correspondng graphs and Table 2 summarzes data assocated wth both the temporal and the spatal overflow evaluaton. These results wll be used n developng and evaluatng the MU SL-BM program based on the SL-RAT Month Movng Avg. 12 Month Movng Avg Lnear Regresson +/-50% onfdence Year Fgure 9 Both short tme and long tme movng averages of the Overflows / 100mles normalzed to a one year tme nterval based on the MU SSO data for FY 00 to FY 12. Table 2 MU Overflow Temporal and Spatal Data Summary Based on FY 00 to FY 12. Entre ollecto n System Rego n 0 Rego n I Regon II Rego n III Total number of overflows Number of square mles Number of lnear mles of ppe lne Overflow/100m rate of change (lnear regresson slope, Fgure 9 and 10) Overflow/100m at November 2011 from lnear regresson Fgure 9 and

15 Overflow Rate / 100m / Year Overflow Rate / 100m / Year Overflow Rate / 100m / Year WEFTE 2012 Sesson 114 O&M of ollecton Systems Month Movng Average 12 Month Movng Average Lnear Regresson +/-50% onfdence (a) Year 3 Month Movng Average 12 Month Movng Average Lnear Regresson +/-50% onfdence 20 (b) Year 3 Month Movng Average 12 Month Movng Average Lnear Regresson +/-50% onfdence (c) Year Fgure 10 Both short tme and long tme movng averages of the Overflows / 100mles normalzed to a one year tme nterval based on the MU SSO data for FY 00 to FY 12 (a) Regon I 0<Overflows/SqM 10; (b) Regon II 10<Overflows/SqM 20; (c) Regon III Overflows/SqM > 20.

16 Locaton (mles) % Ppe Inspected wthn Square Mle Relatve Occurance WEFTE 2012 Sesson 114 O&M of ollecton Systems MU Acoustc Inspecton Evaluaton Over the past two years MU has conducted well over three thousand SL-RAT acoustc nspectons of ppe segments. Ths has provded a condton assessment for over 850 thousand feet of the collecton system. In the past year, the focus of the nspectons has changed from evaluatng the SL-RAT technology, to ncorporatng the technology as the enabler for a newly evolvng collecton system cleanng BM program based on acoustc nspectons. Fgure 11(a) provdes a spatal representaton for the SL-RAT acoustc nspecton locatons conducted by MU. The spatal representaton s based on a one-square-mle grd system; the same system used n evaluatng the overflows n Fgure 8(a). For each one-square-mle locaton, the percentage of ppe nspected by the SL-RAT over the past two years s ndcated. The nspectons ncluded n the evaluaton are restrcted to those nspectons conducted pror-to cleanng operatons. Ths ncluded almost three thousand measurements. By comparng Fgure 11(a) wth Fgure 8(a), an mportant observaton s that the SL-RAT acoustc nspectons were conducted predomnately wthn ether moderate or hgh overflow locatons,.e., Regon II and III. Usng ths observaton, then the SL-RAT Blockage Assessment provdes an estmate for the cleanng requrements wthn the MU collecton system based towards locatons whch are hstorcally more prone to overflows. The correspondng hstogram of the SL-RAT Blockage Assessments s gven n Fgure 11(b). Based on the classfcaton employed by the SL-RAT: 58% of the ppe segments are consdered Good, 19% Far, 14% Poor, and 9% Blocked. The ppe segment condton assessment s an mportant component n developng and evaluatng the SL-BM program. An addtonal refnement would be to evaluate the SL-RAT Blockage Assessment for each regon (Regon I, II and III) ndvdually. To date, the SL-RAT measurement data set does not support estmatng the blockage assessment dstrbuton n each Regon. Therefore, the dstrbuton n Fgure 11(b) s used to represent the three regons sq. m % 14% 19% 58% Locaton (mles) (a) Blockage Assessment (0-Blocked thru 10-lean) (b) Fgure 11 MU Acoustc Inspecton (a) Spatal representaton of the percentage of ppes nspected wthn each square mle grd; (b) Hstogram of the blockage assessment.

17 WEFTE 2012 Sesson 114 O&M of ollecton Systems MU SL-BM Program Evaluaton The hypothetcal SL-BM evaluaton presented n the Overvew s extended by ncorporatng the MU SSO evaluaton presented n the prevous secton and the MU acoustc nspecton program mplemented over the past year. To date, nsuffcent data and nsuffcent tme has elapsed to allow drect evaluaton of the SL-BM program beng establshed at MU; therefore, a performance model s derved based on the emprcal data collected to date. The approach s to evaluate the ost assocated wth the SL-BM program based on establshng a new collecton system cleanng program to acheve a desred number of overflows/ 100m of lnear ppe,.e., the performance goal O T PT (6) NT where O T s the total number of overflows wthn the collecton system based on the utltes mantenance program and N T s the number of 100m lengths of ppe wthn the collecton system. The evaluaton model s derved to evaluate the total cost of the mantenance program T (7) where s the cleanng operaton cost for the th Regon. The value of s evaluated for two cases: leanng Only program wth no Acoustc Inspecton and SL-BM program based on Acoustc Inspecton,.e., where A N [leanng Only no Acoustc Inspecton] (8) I I N A [SL-BM wth Acoustc Inspecton] (9) s the cost to clean 100m length of ppe and I s the cost to acoustc nspect 100m I s the fracton length of ppe wth the SL-RAT. A s the fracton of the th Regon cleaned and of the th Regon nspected wth the SL-RAT. For the SL-BM program, A s determned based on the SL-RAT Threshold used to dscrmnate between ppe segments requrng cleanng and those that do not. The relatonshp between the SL-RAT Threshold and the fracton of ppe segment cleaned, D, s derved based on the MU acoustc nspectons and the relatve occurrence of blockage assessments as depcted n the hstogram n Fgure 11(b). The SL-RAT Thresholds evaluated are the same as those used n the case studes presented above and the values D for the SL-RAT Thresholds are gven n Table 3. Then usng cost for the th Regon s s D I I N Next, the total number of overflows s gven by A D I, the SL-BM [SL-BM wth Acoustc Inspecton] (10) O T O N (11)

18 WEFTE 2012 Sesson 114 O&M of ollecton Systems where O s the number of overflows/100m of lnear ppe n the th Regon and of 100m lengths of ppe n the th Regon. N s the number From the graphs n Fgure 10, the overflows/100m for the th Regon can be estmated for the MU mantenance program pror to the effect of the acoustc nspecton. These values are estmated based on the lnear regresson and are gven n Table 2 and are used n evaluatng O. To evaluate O, the new mantenance program performance needs to be evaluated n terms of the former mantenance program. Usng ths approach, the th Regon s overflow/100m s modeled by O R A R F R [leanng Only no Acoustc Inspecton] (12) O R I R F R [SL-BM wth Acoustc Inspecton] (13) where for the SL-BM program t s mplct that the fracton of the th Regon cleaned s A D R s the rate of change n the number of overflows/100m based on the change n the I. fracton of the area mantaned (cleaned or nspected). F s the fracton of the th Regon cleaned under the former mantenance program and R s the overflow/100m n the th Regon based on the former mantenance program. Table 3 Parameter Values Used n Evaluatng the SL-BM Program ost-performance Parameter Fracton of ppe segments cleaned, D, for SL-RAT Threshold 1 Fracton of ppes segments cleaned, D, for SL-RAT Threshold 3 Fracton of ppes segments cleaned, D, for SL-RAT Threshold 5 Value Rato SL-RAT acoustc nspect cost to cleanng cost, I / 0.09 Rate of change n the number of overflows/100m to the change n the fracton of the collecton system area mantaned, R -7.8 R s an mportant parameter n evaluatng the effectveness of a mantenance program. It specfes the rate n achevng the performance goal based on ether cleanng more ppe segments or by mprovng the selecton process for targetng cleanng resources to the ppe segments requrng cleanng. R s estmated usng the slope from the lnear regresson from the scatter plot data depcted n Fgure 1. For the results presented n the paper, the value s consdered a

19 WEFTE 2012 Sesson 114 O&M of ollecton Systems constant. Ths provdes a frst order approxmaton. R s lkely to be regon dependent and dependent on the mantenance program followed,.e., for the SL-BM program, R wll be mpacted by the SL-RAT Threshold selected. As the SL-RAT Threshold s ncreased, the number of ppe segments scheduled for cleanng ncreases. In addton, the cleanng targets ppe segments whch are ncreasngly cleaner. Therefore, the value of R wll ntally rapdly mprove wth a dmnshng mprovement as the SL-RAT Threshold ncreases. Ths relatonshp has not been establshed and therefore s not used n evaluatng the results n the paper. The desred model for relatng the total cost n terms of the overflows/100m of lnear ppe s obtaned by combnng equatons (6) through (13) where NT K T T K P K F N R N (14) R R K c [leanng Only no Acoustc Inspecton] (15) K D [SL-BM wth Acoustc Inspecton] (16) c I Lookng at the three terms n equaton (14) provdes nsght nto the model. The frst term, N T K R P T, provdes the head room savngs based on the performance goal beng greater than zero overflows/ 100m. The second term, K F N, s the cost of meetng the Former mantenance performance usng the new mantenance program and the thrd term, K R R N, s the cost of mtgatng the reported overflows based on the new mantenance program. By relatng equatons (15) and (16), the mechansm for achevng substantal cost savngs usng the SL-BM over the lean-only program can be readly evaluated. For the same performance goal, the SL-BM wll be less expensve than the lean-only program gven c I c c D I 1 D (17) As dscussed n prevous sectons, the cost of acoustc nspecton s, conservatvely, less than a tenth the cost of cleanng. In addton, the fracton of ppe segments not requrng servcng ( 1 D ) are at least 50% and often sgnfcantly greater. The nequalty n equaton (17) s well met, leadng to substantal cost savngs for the same performance goal. The total mantenance program cost, T, s evaluated usng equatons (14) through (16) based on varyng the performance goal, P T. Graphs of the evaluaton are depcted n Fgure 12. The cost n Fgure 12 has been normalzed by the estmated cost for cleanng 20% of MU s collecton system ( 0.2N T ). Ths normalzaton removes the uncertanty assocated wth cleanng cost,. The parameter values used n evaluatng the equatons are summarzed n

20 ost Rato (BM ost)/(urrent ost) WEFTE 2012 Sesson 114 O&M of ollecton Systems Tables 2 and 3. Three SL-BM programs are compared based on usng dfferent SL-RAT Thresholds. These results are compared to the program based on leanng-only. From Fgure 12, a SL-BM program usng an SL-RAT Threshold of 3, results n a performance of four overflows/ 100m wthout ncreasng cost. To acheve the same performance wth the lean-only program requres over 3 tmes the cost SL-RAT Threshold 3 5 lean Only Over 3x ost Savngs ~2x Performance Improvement urrent Performance urrent leanng ost Overflows/100m n One Year Fgure 12 ost versus performance evaluaton for the SL-BM. ost s evaluated based on the rato between the cost of the SL-BM program wth the cost of the current cleanng program,.e., leanng 20% of the ollecton System/Year. Performance s based on the number of overflows per 100 mles n a year. SUMMARY & ONLUSIONS In summary, we have explaned the basc operatonal prncples of a new ppe nspecton technology and shown through several examples from harlotte-mecklenburg Utltes, (MU), multple ways that actve acoustc nspecton can sgnfcantly mprove the mantenance cost and performance of gravty-fed wastewater collectons systems. Ths patented technology s emboded n a devce called the Sewer Lne Rapd Assessment Tool or SL-RAT whch has nspected over 850 thousand feet of ppe durng the past two years of feld use. The underlyng novel technology tself was descrbed to make the prncple of operaton clear. The SL-RAT s composed of two components a transmtter and recever whch are each placed n adjacent connected manholes. The transmtter generates sound waves that naturally couple nto the ppe segment. The sound waves propagate n the ar gap wthn the ppe segment. The recever measures the acoustc plane wave receved from the transmtter and, through a propretary algorthm, evaluates the condton of the ppe n the form of a blockage assessment. Ths blockage assessment s an aggregate measure of the blockage level wthn the ppe and s scored on a relatve scale of 0 to 10. The blockage assessment scale s emprcally correlated wth TV where zero means fully blocked and ten means the ppe s fully clear.

21 WEFTE 2012 Sesson 114 O&M of ollecton Systems Addtonally, multple operatonal benefts of the SL-RAT relatve to exstng alternatves were covered. The SL-RAT does not requre confned space entry and does not contact the wastewater flow makng t safer to operate. It also does not requre the support of cleanng equpment, provdes the blockage assessment n 3 mnutes or less, has been practcally operated n a typcal wastewater collecton feld envronment under a varety of condtons, and can be easly operated by a feld crew of two operators. The results of multple plots and feld studes conducted by MU n conjuncton wth Infosense were revewed whch show the effcacy, the economcs, and the operatonal advantages of the SL-RAT devce. The effcacy of acoustc nspecton technology was hghlghted n a study conducted by MU and InfoSense to correlate TV vdeo wth the aggregate blockage measurement provded by the SL-RAT. Acoustc nspectons were shown to successfully detect blockages wthn a ppe segment and to provde acceptable resoluton for delneatng when ppe cleanng actvty should take place and when t should not. Ths same feld study estmated the cost of operatng actve acoustc nspecton equpment and found that the SL-RAT s relatvely low cost of operaton combned wth detectng where the hgh cost of cleanng ppes can be avoded can generate savngs n the range of $1.07 to $1.13 per foot nspected. Fnally, we looked at extendng the use of acoustc nspecton technology to enable the establshment of a ondton Based Mantenance (BM) program for gravty-fed collecton system ppe mantenance. An example was llustrated usng data from MU to extrapolate that for a system targetng 2 overflows/100 mles/year, mplementng an SL-RAT-enabled BM program could reduce cleanng costs by 50% or more. The mprovement compounds as more resources are shfted to the relatvely cheaper task of acoustc nspecton and away from the relatvely expensve and partally wasteful task of scheduled ppe cleanng. The fnancal beneft comes through better focusng cleanng crews on blocked ppes and away from cleanng ppes that do not need cleanng. These results were extended further to develop a performance model whch llustrates mathematcally that usng SL-RAT data as part of a BM program can provde sgnfcant benefts to wastewater system operators by producng BOTH a sgnfcant postve mpact on overflow performance as well as system mantenance costs. REFERENES harlotte Mecklenburg (2011), Wastewater Performance Report July 1, 2010 June 30, Howtt, Ivan; Fsburne, John; Wlson, Bran, (2010) Actve acoustc methodology for detectng sewer lne obstructons, N-AWWA Annual onference, November Howtt, Ivan, (2009) Montorng systems and methods for sewer and other condut systems, Pat. Pendng, Applcaton No. 12/399,492, Lehtonen, Matt (2006), On the optmal strateges of condton montorng and mantenance allocaton n dstrbuton systems, 9th Internatonal onference on Probablstc Methods Appled to Power Systems, June 11-15, Moubray, John, (1997) Relablty-entered Mantenance 2 nd Ed., Industral Press Inc. Phlp, Morse M., Ingard, K. Uno, (1968) Theoretcal Acoustcs, Mcgraw-Hll Inc. Wseman, Murray; Ln, Damng; Gurvtz, Naaman; Dundcs, Marton; The Elusve P-F Interval,