CA1050615A - Method for detecting moisture in multiple layer roofs - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
A method for detecting the presence, location, and concentration of moisture in multiple layer built-up roofs.
A plurality of spaced points are first marked and located on the roof to be tested. The relative dielectric constant of the roof at each of the spaced points is then measured, the measurements are recorded, and each measurement is associated with the location of the point at which the measurement was taken. The magnitude of the relative dielectric constant of the roof is proportional to the relative concentration of moisture in the roof covering, thus allowing the points at wet portions of the roof to be distinguished from points at dry portions of the roof.
statistical methods may be employed to better separate the measurements taken at dry points on the roof from measurements close in magnitude which are taken at wet points on the roof.

Description

~0~6~L~
METHOD FOR DETECTING ~IOISTURE IN MULTIPLE LAYER ROOFS

This application is a divisional appllcation of application Serial No; 235,986, filed September 22r 19750 (1) Field of the Invention'. This invention per-tains to methods for detecting the presence, location, and con-centration of moisture in roof coverings.
~ 2) Description /of Prior Art: ~arge industrial and commercial buildings often have flat roofs which require a cover-ing for protection from the elements. This covering often con-sists of multiple layers of tarpaper or felt which are bonded together, usually by covering each layer with hot tar or bitumen before emplacement of another layer. A layer of heat insulating material is commonly provided between the structural roof and the multiple felt layers.
Over a period of time, natural weathering forces re-sult in a deterioration of the integrity of the roof covering.
Alternate heating and cooling of the roof causes cracks to appear in the tarpaper or felt. Moisture seeps between the layers and expands by freezing during the winter and evaporation during the summer, with consequent further separation and cracking of the layers. Eventually the covering deteriorates to the point that water leaks into the insulation and through the roof, necessi-tating replacement of the damaged portion o the covering.

~`! ~

, ' " ~ ~ -. ' : ' :
:. '' , ': :
'':''' : ' ::
:':: - :
~`''''. ' ' ' ' ' ' ' ' ' '' : `` , : . ~

:~5~
For purpo~ of ConVQnienCe in lllustr~tion, the deterloration of multiple layer roo~s in wet are~s ~ay b~ ~, clas~ified a~ flr~t stage penetrat:ion~ ~econd stage pen~-tratlon, and ex~ensive water penetration ox third ~tage S penetration. First ~tage penetrat:ion 1~ the initial phase of water penetration into the roGf. I'h~ top lAyers of felt may be ~omewhat porous from asphalt ~d tar breakdown from normal weathering, or water may be found between two layers arrivlng by capillary action from a surface imperfection. There are no leaks into the build~ng at th~s stage of detexioration, but roo~ damage -~.
may continue at a xapld rate unlass th~ ~ituat~on i8 cor- ~' rected and water entry poin~ ~ealed. S~cond stage pene-tration i8 similar to first ~tage penetxation ~xcept that deterioration has progressed into the lower layers of rooflng elts. Water may now be sandwiched between thr~e or our elt layers instead of ~u6t the top ~wo as in ~irst stage penetration. While ther~ 8t~11 may be no water leak-age into the building at the ~econa staqe, the proteotive layers of ~he roof are in-very bad shape and leaka~ into the ~uilding can be expected at any time. The felt~
bitumen layers must be raplaced. The insulation layer in mo~t ca~es can be ~alvaged, but should be inspected to determine it~ condltion~ At th~ third or exten~ive watex penetration s$age, water ha~ penetrated all tha pro~eotive ~5 felt layers and the insulation a~ well. Since tha binders u~ed $n mo~t in~ulatlon are water soluble, the roof water often dissolves the binder~, and the insulation deteriorates and becomes muehy and also looses its h~at ~n~ulation ability.
In thi~ third ~tage of exten~ive water penetration, both _ -the felt layers and ~he ur~clerLyirlg in~ulatlon are ~oaked, and the ~ntlre roo covering mu~t be xemoved and rebullt -from the ~tructural roof on up u~ing new materials.
In gen~ral, first 6tage penetration area~ can be 5 most easily r~palred by top-coating with a cold applie~
roofing maEtic or an ela~tomaric Coat~ng ma~eriAl . If the penetration i~ indtcated a~ com~ng from a ~la~hing crack or tear, th~ repair should be made with an ela~to-meric material applied over a poly~thylene me~h tape.
Asbestos fillad bituma~tic m~erlals work for temporary repalrs. Thes~ ar~ not ~atis~actory ~or long-term repalrs:
becau~e they become brittle and cracked ~o again allow~
water entry.
Second stage penetration ~ndicates much more 6evere roof deterioration. Only temporary repair3 ca~ be made with the top layers. Permanent coxrective repair~ fox ~econd 8tag9 penetration areas require~ the x~mo~al of existi~g felt plys down to the insulation to completely remove all trapped water. Where vi8ual in~pection indicates excesslve deteriora-t~on of the insula~ion~ it should be removed. The roof~hould be rebu~lt from the lnsulation on up, or a~ required ~rom the decking on up, u~ing standard roofing techni~ues and new dr~ mater~als. Hot tar and hot a~phalt felt laminat~
ara ~a~is~actory here becau~e all roof water has ~een rèmoved.
Hot xoofing materials do not work well over ar~s ln a r~o~
containing moi~ture, and therefore ~hould not be used ln these area~. The repair area~ are built up to the level o~ the axisting ,older; bullt up roo~lng, and ~roperl~ over-~ 5~lapped. Cold procas~ roof ma8~ or el~3tom~rlc coa~ing~
are applled a3 a flnal protective top coat.
Generally, roof coverlng doe~ not detariorate ov~r an ~ntlre roof at once, and repair or r~placement of the entire covering i9 not justifled. Ilo~ever~ it may be dif-flcult to dein~ tha areas of the cov~rlng ~hat need repair or replacement, ~ince water may txavel a con~iderable di~-tanee between ths layer3 to the point where the leak in the roof appear~. The traditional method of determinlng areas of we~nes~ in roof covering~ has been to tak~ core sample~ of the covering. Thi~ method 18 time ~on~uming and ha~ tha obvious disadvantage o~ xi~king destructlon of sound rooE covering if too great an area is te~ted, whlle deteriorated covering far from the polnt of leakage may be missed if extensive samples ar~ not takan. More~
over, core ~amplinq is an impractical method for detecting the f~rst ~tages of roo~ deteriora~ion, which a5 noted above can o~ten b~ repaixed without replacement of the ao~erlngO
Several non-destructiva roof wetn~ss testing technique~
have been de~eloped in an a~t~mp~ to reduce the time and expen~e of testing roofs. The~e technlques include in~ra-red scanning o roof~ to detect cool moist area~, and nuclear particle bombardment to determine hydrogen ion coun~ in the roof covering. Such techniques generally requ$re expen~lve and delic~te equipment, and do not di~-crimlnate the-flr~t stage and ~econd stage penetration wet areas from dry area~ with as hi~h a degree o~ accuracy a~
is de~irable~

~o~ s SUMMARY OF ~'~IE INVENTION
I have invented a new method for detecting the pres-ence of moisture in a multiple layer built up roof, and for deter-mining the location and the relative concen-tration of this mois-t-ure in the roof. My method is capable of distinguishing the areas of first stage, second stage, and third stage moisture penetration into the roofs, thus acilitating the most efficient and economic-al repair of the roof. The method does not require the use of complicated equipment, or highly skilled operators, yet the re-lative concentration of moisture in the roof is measured to ahigh degree of accuracy.
In applying my method, a .number of spaced points are first established on the area of the roof to be tested. These points may be located by laying out a number of spaced lateral and longitudinal intersecting grid lines. The intersection points of : the grid lines form positions in a matrix, each intersection point on the roof being locatable as a position in the matrix. For con-venience the matrix may be reproduced in scaled down form on a sheet of paper prepared by the operator. I have found that the moisture concentration in built-up roofs is proportional to the dielectric constant of the roofing material. Thus, by measuring the dielectric constant at each of the grid line intersection points r it is possible to determine the relative concentra-tion of moisture in the roof at the various points spaced over the area of 1 the roof~ ~t is not necessarv to measure the absolute dielectric - constant of the roof, since it is the relative concentration of moister in the - ~ , ;;~ - 5 -, :
,., ~05~

wet portions of the roof as compared to the dry portions of the roof that is most significantO I have found that the relative dielectric constant of the roof may be effectively - 5~ -. . _ - .
.
.
:

.
: . . . . ~ .

m~asured u~ing a spray field ~apacitance meter, in which the plates of th~ cap~citor are eo-planar. Tha me~urements 50 obtain~d are recorded and are asslocia~ed with the polnt ln the matrix from which they were taken.
In applying my m~thod~ th~ measurement~ obtalned from portlon~ of the roof having fir~t ~age pen~tr~-ion may be quita close in magnitude to the normal ~catter of readlngs ~rom the dry roof portlons. The measurements indlcating first stage penetration may be separated ~rom the dry portion mea~urements by several technique~, including taklng core sample~ ~t a number oE the spaced point3 on the roof and examining the core ~amples to determlne i they ara from wet or dry portion~ of khe xoof. The smalle~t reading obtained at ~oints having wet cora samples can be used to des~gnate all points havlng mea6urements hlgher than such ~malle~t measurement a~ wet poi~ts. Similarly, all point~
having measurement~ smaller than the large~t measurement at a dry poin~ can ba designated as dry poin~s on the roof~
Since mo~t multipl~ layer roofs are similar in construction, and ~hu~ have ~imilar dielec~ric constant properties~ it ls also possible to use the measurements obtained ~rom ~ ~econd roof which is known to have wet portiona and dry portion39 as a ~tandard again~t which the mea~urement~ obt~in~d ~rom the xoof being tested can be compared.
~urther ob-~ect~, feature~ and ad~antage~ of my lnvention will be appa~ent from the following data~led description tak~n in con~unctlon with the accompanying drawings ~howing a pr~ferred methocl for detecting molsture in multiple layer roof~ exempli~yi~g the prinoiple~ of my inventlon.

~s~
BRIEF DESCRIPTION OF THE DRAWINGS

FIGURE 1 is a perspecl:ive view of a building having a multiple layer roof with a roof capacitance measuring unit in position to take measurements thereon.
FIGURE 2 is a scaled drawing of the roof of th~ build-ing shown in FIGURE 1.
FIGURE 3 is a cross sectional view of the roof of the bullding in ~IGURE 1, with a capacitance meter in position there-on to take relative capacitance--relative dielectric constant measurements.
FIGURE 4 is a graph showing the frequency of the re-lative capacitance readings taken on the roof of the building shown in FIGURE 1.

DESCRIPTION OF A PREFERRE:D EMBODIMENT

Referring now more particularly to the drawings, where-in like numerals refer to like parts throughout the several views, a roof capacitance measuring unit used in my method for detect-ing moisture in multiple layer built-up roofs is shown generally at 10 in FIGURE 1. The capacitance measuring unit 10 is shown in FIGURE 1 on the substantially flat roof 11 of a building 12 such as an industrial plant or commercial facility. The capa-citance measuring unit consists of a capacitance meter 13 elect-rically connected by a cord 13a to an ammeter reading dial unit 14, both o~ which are mounted for convenience on a wheeled cart 15.

.', ; ~ - 7 -"

.s Referrinq to FIGURE 3 the capacitance meter 13 is shown in operating position on a cross section of the roof 11.
For - - 7~ -....... .
.. . .. . .
, . . .....
' ' .

. .

~5~
illustrative purposes, the roof ]1 i5 shown as having a smooth surface. However, the presence of gravel, pebble~, etc., on the surface o~ the roof does not affect the application of my method.
The roof 11 is composed of multiple layers of felt or tar pap~r lla which are bonded together by an adhesive layer llb, this adhesive usually being bitumen or tar. The layers of felt lla and tar llb are layed over an insulation layer 11¢ which is on the structural roof lld of the building 12. The structural roof lld is shown for illustrative purposes in FIGURE 3 as composed of concrete, but my capacitance method for detecting moisture can also be used with roofs having any other structural material such as steel or aluminum. Although a conductive structural decking may increase meter readings, the increase in readings is subs~antially uniform in the range of readings near the dry roof portion rPadings, so that wet areas will still yield relatively higher readings as compared to the dry areas. As described above, the layers of felt lla are capable of absorbing water where cracks or separations have occurred in the felt and tar layers.
Water may also accumulate between layers, causing further layer separation because of freezing and evaporation forces. As a xe-sult, water will seep through the protective layers of felt and tar to the insulation layér llc and will probably leak through the structural roof lld, which usually is not constructed so as to be water-tight. In addition, there are various openings in ~ - 8 -., .. .. :, --` ~

the structural roof such as the skyliyht 16, roof access open-ing 17, plumbin~ vents 18, and drains 19 as shown on the roof 11 in FIGURE 1. The presence and relative concentration of moistuxe absorbed by the felt or insulation, 8~
,. .,~ :
:
:::
'~, ,. . ~ , .
`` ', ' ~ :
: ' ~

`' ~IL~5~
or between tl~ layer6 i9 detecte~ by the capacitance meter 1~ a~ explalned belo~. It i8 important that the xelative conc~ntra~on of moi~ture ln the layers b~ de~ermlned b¢cau~
those areas o ~he roof which have first or econd ~tage moisture penetration may need to be replaced or repaired even though not presently vi~ibly cracked.
. The capaci.tance meter 13 prefe~ably utili2e~ two co-planer ~lectrode~ 20a and 20b which ar~ moun~ed ~t the bottom of the meter 13. The~e electrode3 are ele~trically ~olated from one anothar, and are connected to a source o~
constant fr~quency alterna~ing current ~AC~ powerO A ~pray or fringe electro-~tatlc fleld will exl3t between the two electrode~ and ~hey will in effect form two plates of a capacitor, with a small amount of electric current flowing through the capacitor~ Thi3 cuxrent will be measured by the ammeter xeading dlal unit 14. Th~ amount of current flowing in the circuit will b~ proportional to the capacltance of the - ~ two electrode~ 20a and 20b, which i8 in turn proportional to the dielectric constant of whatever materlal i8 in the spray ; 20 field between these two el~ctrodes. By de~in$tion, a ' dlélectric mat~rial b~tween the plates of a capac~tor increases the capacitance thereof over the ~n-~acuum capacitance by multiplying the in-vacuum capacitance by the dielec~r~c cons~ant of the material. I have found that the ~ 25 diel~ctr$c constant of falt and tar layera in a buil~-up ; roo in~raa~e~ wlth an increa~e ln the conc~nt~atlon of ~ ; : moisture .in the layer~ and between the layers. Thus, it ~s : po~sible to obtain a read~ng o the relativ~ diel~ctric con~tant and hence the relatl~e concentration o moisture in the roo. layers by placing the electrode~ 20~ and 20b of ~ ~S~6~5 the capacitance me~r 13 ac3ain~t ~hs roof ~urface or agalnst the gravel or pebhle~ on the ~u:rface, and xe~ding on the readlny dlal unit 14 the current th~t 1~ flowing in th~ circuit~
It i6 to be expectad th~t the dlelectric properties of roofin~ mat~rial will vary ~:rom buildi.ng ~o building, and from polnt to point on the xoof of a sinqle building~
These vari~tlon~ can occur becau~e of diEf~rence~ in the material itself, in the number o~ layer~ of materi~l that have been placed cn the buildin~, in the thlckn~8~ o~ the : 10 roof layer~ and the gra~el layer, s~paration of the layer~, varla~ion~ ln surface temperaturo, and 60 forth, including as . one variable the relative concentration of moisture contained within the layers of roofing material. In addltion, where water i~ trapped or absorbed only in a single layer or in very deep lS layers of roo~ing materlal~ the readings tha~ are obtalned . . may not vary greatly from the normal ~cattQr of readlngs o~ained from dry roof portions. However, the presence of moisture in a ~ingle layer or in deep layexs must be ascertalnsd ~ince the roofing wlll eventually begin to deter~orat~ in .the~e ar~a~. It is po~sible, by us~ng my method, to.
- dlscrlm~n~te b~tween th~ dry portion~ o~ the roof and the wet portions wh~ra first or se~ond stage moi~ture penetration has occurred, as will be developed below.
:: In ~pplylng my method o de~ecting moisture i~ roofs t the roof 11 ~o be measured ls fir~t marked o into a grid or matrlx whi~h~E)referably ha~ uniformly s~a~ed 1nter~cction polnt~
- at the corner~ of grid ~quare~ that are approx~mat~ly 5 feet . on a side. The later~l and lon~tudinal intersecting grid lines .

)~

; ''' ~

~ ~ 5~ 6~ ~
21, as shown in l'IGURE 1, may be phy91c~11y marked on the roof 11 of the building hy ch~lking the l.ines on the bulldlng or ~y mar~lng them off with tape, or by ~erely marking the intersection point~ 21a of the gridlines.
A scaled drawi~ of th~ roof 11 with the ~rlcllines 21, as shown ln FIGUR~ 2, 19 then made. Each gridline intersection point is located b~ assigning the point a lateral and longitudinal poqition in the matrlx, a9 (0,0~, ~A,l), ~B,5), etc. in the manner ~hown in F~GU~E 2.
The measuring unit 10 i~ then placed on the roof 11 and the capacitance meter 13 is call~rated ~o a conV~nien~ value with the capacitance meter in the air away from any di~lectr~c mat~rialO Thus the measuring unit will not measure the absolute dielectric constant o~ the roof coverin~, but will read the dielectric constant relative to air, or for that ma~ter, relative to any other oonvenient mater1al~ Alr provides the most convenient reference material, BinCe its dielectric properties are airly constant and close to that of a vacuum, and it is availahle anywhere. The operator then pu.~h~s ~he measuring unit 10 along the gridlines 21 and 8tops to take relative capacitance or rslative dielectric constant mea~Uremen~s with the meter 13 at each gridline intersection point 21a. Ile marks down the value of the relati~e d~lectric constant re~d~ng as shown on the 25 reading dial Uilit 14 on the scaled drawing shown in FXGURE 2, ; wi~h each rè~ding b~ing marked next to the intersection poin~ 21a which is associ~t~d with that reading.
Alternatlvely, the drawing of the roof 11 with the gridlines 21, as shown in FIGURE 2, may be prepared initially . . .

.. . .. . .
. . .

~ror~ measurement~ taken o~ the cllm~nslon6 o~ the roof. The actual rea(lin~s with tl-~ measuring unit 10 may then b~ matle by pushing ~he unit alon~r a st:raisht llne and taking measuremenk~3 every S feet, and continulng ~chls pattern back and forth across the roof until readings hEIve been taken over ~che entire roof. 1~ flve-fvot square gr~.d has been chosen front experience as yielding the maximum probabili~y of detecting wet areaR o~ the roof, while mlnimi~ing the amount of work required to t~ke measurements over the Qntire roof area.
'rhe readings which, for illustrative purposes, have he~n marked ~o the upper right of the inter~ectlon points of the gridlines on the drawing in FIGURE 2, were obtain~d with a par~icular model of capac~tance meter 13 calibratecl to a p2rticular reading ln air, wherein a capacitance raadln~
in air would he approximately " O " . The relativ2 capacitan~e < readings markea on th~ drawlng in FIGUl?E 2 ar~ collected and shown in graphlc form in the bargraph of FIGU~E 4. A~ shown - in FIGURE 4, the relative capacltance readlngs t~nd to cluster ~ 20 around a xeading of 5 or 6~ with only scattered r~adings .. ,~ .. , :. . ....... .... . . . ....... . . .............. . .............. .
occurrlng out at higher valu~s~ It has been found lthat wi~h the particular mcdel of capacl~cance meter 13 calibrated in alr as above, that dry roof ~ generally tend to read between 3 or 4 and 6 or 7. It has al~o been found that the - 25 standard doviation of rQadings 'caken on dry roofs seldom.
~ . .
exceeds.l ~r ~ where the meter 13 ~ ~1mllarly calibrated '' in air. ~lëncQ, read$ng~ of more than 12 or 13 are almo~t ~urely not readings taken at dry portion~ of the roof.
The mean and standard deviation o all readings excluding r. . ~ ... . - -, . ..

: ' ' .

.. . .

6~S
those readin~s over 13 can then ~be calculated. The standard formulas for the sample mean M and sample variance S are, of course, M ,- 1 n - ~ Xi and s2 = 1 (xi-M)2, where x ~ i-l is the value of readiny "i". The sample mean and sample stand-ard deviations so calculated are considered the estimators of the mean and standard deviation of the population samples from dry portions of the particular roof being measured. There is thus a high probability that any reading that exceeds the dry roof mean plus 3 dry roof standard deviations is a reading taken at a portion of the roof that is wet. Readings beyond 14 or 15 can be conclusively assumed to be at wet portions of the roof.
For the readings shown in FIGURE 2, the dry roof sample mean is e~ual to 6.75 and the dry roof sample standard deviation is equal to 2.135, and thus the sum of the sample mean and three times the sample standard deviation is 13.155, which agrees with the origin-al estimate.
The concentration of water in the layers can vary great-ly, and water may be contained in one of the deep layers but not in the upper layers, or in fact in more than one layer. The readings taken at wet areas of the roof will thus tend to vary greatly and will not cluster around a single value, but rather Il~
;- ~ - 13 -; . .:. : . .

- :. : , ,, ~ .
.,, . ... , , : :
' :: . . . : . ' ,. ', . :
.~ ,, . ~ . .
. . . . . .

~5~5 will cluster around several reading ~alues. The reason for such clustering can be s~en from an examination of Tables l-S below.

'.~, .
.

, . .

:, .

5~96~Si Table 1. Effect of Layer Position on Meter Readings for one Wet Felt Layer and Dry Insl~r~tion (Meter reading on dry insulation ~ 3.0) Meter Readings . ,_ __ _ _ _ _ _ Wet Fe:Lt Wet Felt Wet Felt - Number of Wet Second Third Fourth Felt Dry. Bottom From From BottomFrom Bottom Layers Felt Felt Bottom 1 3.5 7.8

2 4.0 7.8 7.8

3 4.3 7.7 7.6 7.8

4 4~5 7.7 8.0 8.g 8.1 4.6 7.7 8.0 8.0 8.2 i : Table 2. Effect of Layer Position of Meter Readings : for two Wet Felt Layers and Dry Insulation (Meter Reading on dry insulation = 3.0) . Meter Readings : __ ..
Wet Felt Wet Felt Wet Felt ` A
. Wet Second Third and Fourth : Number of Bottom and Third Fourth and Fifth : Felt Dry Two from from from Layers Felt Felts .Bottom Bottom Bottom :' . __ _ 1 3.5 2 4.011.0 ~:~ 3 4.310.5 11.0 . - 4.510.~ 10.~ 11.4

5 . 4.610.0 10.3 10.9 10.5 ` .

:, .
. - 14 -: . , :'. : ' ` : ':` :
.::~ . : : : , .
:, , Table 3. Effect of Layer Positlon on Meter Readings for three Wet Felt Layers and Dry Insulation (Meter Reading on Dry Insulation ~ 3.0) Meter Readings ,., , ~ ~

Number of Dry Wet Bott:om Wet Felts Second Felt Layers Felt Three Felts Third and Fourth from Bottom .
1 3.5 2 4.0 3 ~.3 13.0 4.5 1~.1 11.4 4~6 11.6 10.5 Table 4. Effect of the Number of Wet Felt Layers on Dry Insulation (Meter reading on dry insulation =3.0) Number of Wet Felt LayersMeter Readinq 1 8.0 2 11.0 3 12.5 4 13.0 13.1 - ~ ~ 15. _ .
, -': , , . .: ' ' , .

.~' .

~56~6~5 Table 5. Effect of the Number o~ Wet Felt Layers on Wet Insulatlon (Meter read.Lng on dry insulation = 3.0) Number of Wet Felt La~ers Meter Reading 0 16.0 1 18.0 2 19.5 3 20.1 4 . 21.0 21.5 '''' ' - 1 ~ -:: . :. : .- . . - -:-:
: ~ ~ . ' ... :' , .: :: : : .
: , . : . . . , ::

Where one felt only is wet, the meter readings (relative dielect~ic constant measurements) tend to cluster around 8.0, as shown in Table 1. Where two or more felts are wet, the meter read-ings tend to remain within the ranye from 10.0 to 13.0, as shown in Tables 2-4. Large meter readings of 16.0 or more are obtained when the insulation layer is wet, as demonstrated in Table 5. It is thus seen that for the type of roof being tested here, readings clustered around 8.0 indicate stage one penetration~ readings with in the range of 10~0 to 13.0 indicate stage two penetration, and readings over 16.0 indicate stage three penetration Thus it would be difficult, if not impossible, to esti~
mate the sample mean and sample standard deviation of only the population of readings from wet portions of the roof that are close to the dry roof portion readinys. Certainly the parent popula-tion of all wet roof portions would not be normal, since the pop-ulation obviously has several modes. It is, of course, desirable to ascertain as many of the wet roof points as possible without including therein, as error, an unreasonable number of dry roofing points. It may be assumed here that the dry roof readings are taken from a population that is approximately normal, which cor-responds to previous experience. The mean of this population can be estimated by the sample mean M of 6.75, and the population standard deviation can be estimated by the sample standard devia-tion S of 2.135. If such a normal population is assumed, only 10%
of the dry roof readings should be greater than 9.5 and only 5% of the dry roof readings should be greater than 10.27. Thus, if all ~ - 16 -::. . .
'~

of the xeadings which are greater than 9 are con~idered to be wet readings, at a maximum it is to be expected that only -i - 16~ -: .
.:
. . ..
- . ., . ~ , :

,' ' ' '` ' .
.

5~6~
10~ o~ the actual dry roof readings would be mistaken for wet roof readings~
This error estimation m~thod can be generalized to re-duce the maximum probability of e.rror (i.e~ designating a dry point on the roof as a wet point) below any desired probability P~ The relative dielectric constant measurements taken at several points which are at wet portions of the roof are first determined. This may be done in several ways, including taking core samples at several of the spaced grid intersection points and determining from inspection of the core samples whether they are from wet portions of the roof. Where a wet core is found, that grid intersection point from which the core is taken is designated as a point at a wet portion of the roof. A second method of determining several wet points utilizes the experience gained from measurements on other roofs having similar construction to the roof being tested.
Construction of two roofs is similar if both roofs have coverings having substantially the same number of layers of the same or similar material. Most multiple layer roofs have at least four layers of felt and bitumen over insulation and are thus substan-tially similar, although the construction of the underlying struc-tural roof may vary. However, as indicated previously, a conduc-tive structural roof may add a substantially uniform increase in readings to both dry and wet roof readings in the range near the dry roof readings, and thus the effect of the structural roof can : .

~ ' , ~ ' ' . ;
'.' ` ~

~5~6~
be compensated for. ~s demonstrated in Tables l-S fox the par-ticular .roof being tested there~ with the dry roof readings tend-ing to cluster around a reading of 5.0, wet roof readings should begin at a reading of 8.0, and the roof is almost certainly - 17~ -.. ~ . . -- - .
. ~ , .~ ,. , ' ' ., ' .
'`: ' . ' ' , ,. ~' ' , .:: , .,: . . . . . .
.

wet where readings of 10.0 and above are obtained. Thus~ on a roof similar to that for which the test results are shown in Tables 1~5, relative dielectric constant measurements greater than : 10.0 are almost certainly taken from wet portions of the roof and may be designated as such.
T~e sample mean M and sample standard deviation S are then calculated for all relative dielectric constant measurements which are less than any measurement obtained from the points that have previously been determined to be at wet portions of the roof.
The parent population of dry portion measurements is estimated as having a normal probability distribution with estimated mean M
and estimated variance S2. The actual mean and variance of the parent population will very likely be less than M and S respec-tively, since only measurements which were almost certainly taken from wet portions of the roof were excluded from the calculation of M and S2, whereas it is quite likely that some high measure-ments taken from wet portions of the roof were included in the calculations.
For a random variable having a normal probability dis-tribution of mean M and variance S , there is a number K such thatthere is less than a probability P that the random variable is greater than or equal to K. Given M, S2, and P, the number K
can be calculated from tables of the standard normal distribution, .

~ - 18 -; .' , ' ~ - ' ' " ' : ' ' , :: . - `' : ~ :
, ~ '' ` ', . .

1~5~
using conventional techniques. For the numerical example given above, where P 3 0.1~ M -6.75~ and Sa2.135, K is found to be 9.5.
Thus, any points on the roof having relative dielectric constant measurements whlch are greatex than K can be designated as points - from wet portions of the roof with less than a probability P of designating a dry point as a wet point.

; - 18~ -' .
~:; . . :
~ . , :

.. . . .
.: , :
'' ' : ' ~5~
In order to obtain readings that are even more accurate, it may be desirable to take readings on a day that has followed several days of dry weather to establish a good baseline for dry roof readings. It would then be desirable to take readings again after a heavy rain, or, if this is inconvenient, to irrigate the roof over a period of a da~ or so to insure that a~y wet areas, or areas that are susceptible to wetness, will have become thoroughly wetted.
After measurements have been taken of the relative di-electric constant at each of the spaced grid intersection points,the general areas of the roof where there is moisture in the roof covering are known. However, it is often desirable to obtain a more precise outline of the wet areas than that provided by the grid intersection points. Thus, a relative dielectric constant measurement can be taken at least one point which is between a wet point and an adjoining dry point. If the measurement shows that the point is dry, that point can be designated as an outer limit of the wet area. If the point is wet, another measurement may be made between that point and an adjoining dry point to better define the limits of the wet areas.
It is also desirable to obtain a more precise moisture profile of the wet areas between the points designated as at wet portions of the roof. Relative dielectric constant measurements .~ .
:~' '$ P~- 19 -i, .
: ' :, .' ~.
. : ` , :~5~5 may thus be taken at at least one point between adjoining we-t points. Several of these measurements between wet points are shown for illustrative purposes in E'IGURE 2. This expanded moisture profile allows a determination to be made of those wet portions of the roof having first stage penetration, second stage penetration, and third stage ,~

: - 19~-' ~ ' ' , ~L~S~ 5 penetration, in accordance with the magnitudes of the relative dielectric constant measurements. The nature of the repairs re-quired on the roof in the wet portions can thus be determined.
It is understood that my invention is not confined to the particular methods herein illustrated and descrlbed, but embraces all such modified forms thereof as may come wi.thin the scope of the following claims.

;"., . ', . . .
,; , : ~ . :' :.:: . .

, ' ' :, .
:- ~ .. ..
..

Claims (2)

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:

1. A method of detecting the presence and location of moisture in a multiple layer roof comprising the steps of:
a. locating a plurality of spaced points over the area of the roof;
b. measuring the relative dielectric constant of the roof at each of said spaced points;
c. recording the relative dielectric constant measure-ments obtained at each of said spaced points and associating each measurement with the location of the point at which it was taken;
d. locating a plurality of said spaced points which are at dry portions of the roof by taking relative dielectric constant measurements at a plurality of spaced points on a second multiple layer roof which is substantially similar in construction to the roof being tested and which is known to be substantially dry, calculating the sample mean and sample standard deviation of the relative dielectric constant measurements taken on said second multiple layer roof, and designating as dry points on the roof being tested any points having relative dielectric constant measurements which are less than the sum of said second roof sample mean plus three times said second roof sample standard deviation;
e. comparing the relative dielectric constant measure-ments taken at said spaced points over the area of the roof with those measurements taken at said points which are located at dry portions of the roof; and f. designating as wet points those points on the roof having relative dielectric constant measurements which are greater than any measurement taken at said points which are at dry portions of the roof.

2. A method of detecting the presence and location of moisture in a multiple layer roof, comprising the steps of:
a. locating a plurality of spaced points over the area of the roof;
b. measuring the relative dielectric constant of the roof at each of said spaced points;
c. recording the relative dielectric constant measure-ments obtained at each of said spaced points and associating each measurement with the location of the point at which it was taken;
d. locating a plurality of said spaced points which are at dry portions of the roof;
e. calculating the sample mean and sample standard deviation of the relative dielectric constant measurements taken at said points which are located at dry portions of the roof;
f. designating as wet points those points on the roof having relative dielectric constant measurements which are greater than the sum of said dry portion sample mean plus three times said dry portion sample standard deviation.