Coastal Engineering, 11 (1987) 219-289 29 “lover Selene Publishers BV, tsterdatn—~ Printed in Tho Netherlands Stability of Breakwater Armour Layers — Design Formulae SBNTSIE.W, VANDER MEER Delft Hydrauies, Harbours, Coats and Offhore Division, P.0. Box 182, 6200 AD Emmeleord ‘The Netherlands (Received Jly 15, 986; revised and accepted Mate 17,1987) ABSTRACT Van dor Moe 11:219-298, 1, 1967. Stblty of breakwater armour layers — Desig formule Coastal Ene, ‘Now practical design foremlae have been developed which describe the stability ofrubble non revetments and breakwaters andar endom wave attack. Th forme at based upon a series of rode ests, More than to hundred and Sty teats have been perforned in order to ensure tht ithe rlvant varablea col! be varied apstamaticaly. Aa a rel ofthe present eres of inves. tigations the main shortcomings in Hudson-type formulae have been sled. Stabiiyformloo se given which inelude the lafbenc of wave period, mb of waves, ena eating. pectin shape groupings of waves and the permeability of thecore A clay defined damage ve art ‘erie atrodeed inthe formals INTRODUCTION ‘The use of coarse materials, such as gravel and natural stone for slope re- vetments and breakwater, is very common in civil engineering, In recent years, there has been an increasing demand for reliable design formulae, because of the ever growing dimensions of the structures and the necessity to move into more hostile environments, ‘The Hudson formulas well known because ofits simplicity. In the last de- cade, however, it has been found by many users to have a lot of shortcomings. Tt does not include, for example, the influence of the wave period and does not take into account randem waves. Tho study of Ahrens (1975) in a large wave tank showed the importance of the wave period on the stability of riprap. ‘The tests, however, were performed with regular waves, Evaluation of Abrens' data by Pilarezyk and den Boer (1983) produced stability formulae which included the wave period. Losadaand Giménez-Curto (1979) gave formulae for stability ofrubble mound slopes under regular wave attack which also:included the wave (0378. 3839/87/90050 © 1987 lever Sclence Polishers BY20 period, Hedar (1960) showed the importance of the permeability ofthe struc- ture. His tests were also performed with regular waves. An extensive investigation was performed by Thompson and Shuttler (1975) on the stability of rubble mound revetments under random waves. Ono oftheir ‘main conclusions was, that within the scatter of the results, he erosion damage showed no clear dependence on the wave period, Re-analyzing their data, how- ever, the author has found a very clear dependence om the wave period! Thomp- son and Shutter, infact, used steep waves with a small range of wave periods. ‘Their work has boen used as a starting point for the present research, The dependence of erosion damage on wave period has now been confirmed for a wider range of conditions by performing tests with longer wave periods, The dependency of erosion damage on other variables has also been considered in ‘the present investigetion, ‘The first part of the present investigation has beon described by Van der Meer and Pilarczyk (1984). Stability formulae were given which were mainly based on rubble mound revetments with an impermeable core. Rubble mound breakwaters with a permeable core were subsequently tested later on and reported on by Van der Meer (1985). ‘The last tests of the programme were finished early 1986. An analysis of the complete research programme is given here, This analysis has produced two design formulae which take into account all three hundred tasts performed in the investigation and the results of "Thompson and Shuttler (1975). A design formula for armour units should be # method of determining the minimum mass of individual armour units, for given mass densities, required {or stability as a function of all the variables involved. Inthe following diseu sion the size of armour units is referred to as the average mass of graded rubble, ‘Wey, oF the nominal (cubical) diameter, Daso, where: Dao = (Wea Pa) a and where: yyy = nominal diameter (m) Woo == 50% value of the mass distribution curve (kg) De mass density of stone (g/m?) ‘The relative mass density ofthe stone in water can be expressed by: 4=p,[p=1 @ where:4 ? relative mass density (—) ‘mass density o: water (kg/m®) As shown by many authors there are a large number of variables affecting ‘armour stability. The variables investigated are shown in ‘Table 1. ‘The Hudson formula can be rewritten in a very simple form by using the nominal diameter, D,g., and the relative mass density, 4, as: H,/ADyso= (Ky cot 0)" () The dimensionless wave height, H,/4D,co is the same as the frequently used stability number N,. ‘The wave period can be related to external processes, for example waves breaking on a slope, by the dimensionless surf similarity parameter, ¢, where: Eston a/ On He 7 ay ‘The average wave period, 7, is used instead of the peak period, 7, The use of T. takes into account the influence of the spectrum shape on stability, where ‘thepeak period, 7,, docs not, This aspect is treated later on in this paper. Plots of ./AD oo (oF N,) versus €, are used by many authors to show the influence ‘of wave height, wave period and slope angle on armour stability. ‘The dimensionless camage level, 8, can be described by: S=A/Dioy (8) where A is the eroded cross-sectional area of the profile see Fig. 1. A physical description of Sis the number of eubie stones with a side of Dyan eroded within ‘width of one Dygy. The “no damage" criterion of Hudson and Ahrens is taken to TABLE Primary variables alfeting armour tbiity Vevible Symbol Dimension ‘ominal dame Dow = ralatice masedonsity = Sgniieantwave beght = " ' « s = x ho eet = rowpinesscfwaves = GRY Dermenbiltyofcore = ety mt “DescribdinFig. 1 Damage profile wit erosion area generally tobe when$ is between 1 and 3 and “failure” of the slope is assumed ‘when is greater than 10. The exact value of S for these criteria is dependent to some extent on the slope of the revetment. The advantage ofthis definition of damage is that it is independent on the length of the slope. "HST EQUIPMENT, MATERIALS, PROCEDURE AND TEST PROGRAMME, All tests were conducted in a 1.0-m-wide, 1.2-m-deep and 50.0-m-long wave flume with the test section installed about 44 m from the random wave gen: erator. A system developed by Delft Hydraulics was used to measure and com ppensate for reflected waves at the wave board, With this aystem standing waves ‘and basin resonance were avoided. The incident significant wave height was measured with the structure inthe flume, by means of two wave gauges, placed apart a quarter of the wave length. For the investigation a surface profiler was developed with nine gauges placed 0.30 m apart on a computer-controlled carriage. The surface along the slope vas measured every 0.040 m. Depending on the slope angle every survey con: sisted between 500 and 1600 data points. Successive soundings were taken at exactly the same points using the relocatability of the profiler. An average profile was calculated and plotted by computer and used for determining the erosion damage, Broken stone was ssod for the armour layer, the main characteristics of which sore: Wep=0.128 ke 2, =2680 kg/m"; Dyoo=0.036 m; layer thickness 0.080 m, ‘The sieve analysis curves wore straight lines on a log-linear plot, see Fig. 2. ‘Two gradings were used: Dyy/Dy.=2.25 (riprap) and 1.25 (uniform stones) respectively. The fil-or layer was defined by: Day (armour) /Dyoy (filter) =4.5, land Day/Dyy=2.28, according 10 the tests of Thompson and Shuttler (1975). ‘The thickness of tho filter layer was 0.02 m. When an impermeable core was being tested the filter layer was placed directly on a slope constructed of mor tar, When a permeable core was tested the armour layor was placed directly on the core, without a special filter layer. During the tests with a permeable coreae f Be Tr El 468% % 40 com Fig.2 Sievecurves ota = 8.0 2.20 009 He/A0ns0. Fig. 8. Bxample of daroage aves,am ‘TABLE? ‘Test programme Slopeancle Grading Spectrum Cone Ralaive Nomber| cote DalDy supe permeability ass ofteas Sensity 4 2 235 Pa mone Tas ry 5 235 PM one Ass » 4 235 PM none iss a 6 235 PM none 1s 25 3 15 PM none sz a ‘ 135 PM, none is 0 a 235 narow none as 1° a 225 de one 18s 2 3 15 PM permeable as ® 2 135 PM permeable se 2 1s 135 PM permeable Lez 2 2 135 PM Homogeneous 82 6 2 135 PN permeble 095, 10 2 135 PM Permeable 205 0 PM Piemon Moskowitaspecrum. the grading of the core was: Dyy/D,,=1.50 with Dygo=0.011 m. ‘This means that for the tests with a permeable core, Dygo (armour) /Dyso( core) =3.2. ‘ach complete test considered of a pre-test sounding, atest of 1000 waves, an intermediate sounding, a test of 2000 more waves, a final sounding. After each complete test the armour layer was removed and rebuilt. A test series consisted generally of five tests with the same wave period, but different sig- nificant wave heights. Wave heights ranged from 0.05 in to 0.28 m and wave periods from 1.3 t0 $28. A water depth of 0.80 mm was applied for all tests. A damage curve was drawn for N= 1000 and N=3000, for each test series, as shown in Fig. 8, where N is number of waves, in the test, From this curve the H,/4Dqqq valve was taken for several damage levels, S, and the surf similarity parameter, , given in eqn. (4) was calculated, using the average wave period, ‘T,, ofthe test series. ‘For example, the curve for N=1000 gives Hi/ADyso see Fig. 3. With A= 1.615, Dygo=0.086 m, T,=2.20 sand cot a=3, the surf similarity parameter becomes ¢,=2.97, eqn. (4). The calculated val- ‘ues wore used in IZ,/4Dy29~¢, plots, The damage levels were chosen between start of damage and ‘ailure of the slope (Liter layer visible). ‘The test programme is summarized in Table 2. {64 for the damage MODEL TEST RESULTS Most of the model test results have already been published in the form of H,/4D,sp~ plots by Van der Meer and Pilarczyk (1984) and Van der Meer5 (1985). Due to space limitations these plots are not repeated here. All test results, however, are given in Figs. 9 and 10. Influence of armour layer grading ‘Tests were carried out at slopes with cot «=8.0 and cot a=4.0 with widely sgraded riprap, Das/D,.=2.25, and uniform stones, Dye/Dy,=1.25. Test results are described by Van der Meer and Pilarezyk (1984). The damage to both sgradings was found tobe the same. It can be concluded that the grading of the ‘armour within the range tested has no influence on thestability and that, within this range, the armourlayer can be described simply by the nominal diameter, Das Influence of spectrum shape and groupiness of waves ‘The main part of the prosent series of tests was conducted with a Pierson Moskowitz (PM) spectrum. The test series with a slope angle of cot a'=3.0 was performed with a very narrow spectrum and also a wide spectrum. A com: parison of these spectza (with the same H,and 7.) is shown in Fig. 4.’The test results have been described in detail by Van der Meer and Pilarczyk (1984) and are shown in Fig. 8, Using the average wave period, 7,, for esleulating the surf similarity param: ter, £,—eqn. (4), gave good agreement between the stability test results for a narrow and a wide spectrum, Using the peak period, 7,, the results of the wide spectrum in Fig. 5 wil shift tothe right and large differences will exist between ‘the narrow and wide spectrum. It can be stated, therefore, that stability is not influenced, or only slightly influenced by the spectrum shape or by the grou- ppiness of waves, using the average period, T.. This conclusion was also reached for the development cf gravel besch profiles, by Van Hijum and Pilarezyk (1982). Influence of number of waves In the investigation of Thompson and Shuttler (1975) profiles were sounded after every 1000 waves, up to N=6000. The storm duration showed a large influence on stability, using random waves. Stability is often reached within a short period of time waen monochromatic waves are used, Therefore, a large difforence exists between testing with monochromatie and random waves. Analyzing the resus of Thompson and Shuttler gives the relationship between damage and number of waves shown in Fig, 6. Data points are gen: erally based on 60 tests and are independent of slope angle, wave period and damage level, Since a clear relationship was found betwoen damage and num: ber of waves, it was decided that the total number of waves in the present study206 evenor eseny ns Fig. Specie mith the same Hand could be reduced to N=800 with one intermediate sounding after N'=1000. ‘Analyzing the ratio cf damage after 3000 and 1000 waves and using the damage after 3000 waves as a reference, gave a new point (*) in Fig. 6. The difference between the rosults of the two investigations is small ‘The relationship between damage, S, and number of waves, Nan be described by a squaze root funetion when IV is between 1000 and 7000 waves, as shown in Fig. 6, For this range the influence of the number of waves on stability ean be desesibed simply by the parameter group S//N. "Thompson and Shutter performed five long duration tests with NV upto 15,000, These data are also given in Fig. 6 together with the data of 14 tests with N= 600. ‘A function which describes the influence of the storm duretion on stability completely should moot tho following requirements: — From N=0 to N=500 or 1000 the funetion should be almost linear> 180,50 ee pon symbols: wide spectrum, E94 2020, OF s072 ig 6 fluence of pectrum shape an stability. 160 S«ND / 8(50003 ° 5100 0000 © nore then 60 tests A only 2+ 5 toate 15000 ‘20000 N= Nunber of waves ----- sttorsc6000) ig 6 Innes of amber waves on dase.28 ‘TABLES {ower and upper amagsteves Stopoanale Damage level 8 Onset ct damage Failure 2 5 2 8 2 2 3 a 3 — For large NV numbers there should be reached a limit to the damage (equilibrium). A function which meets this requirements is: f(S) a1 —exp (-6N)] (6) where @ and b are curve-fitting coofficients. Based on the data of Fig. 6 the ‘coefficients, a and b, are found to be 1.3 and 3% 10~* respectively. The infh- {ence of the storm duration on stability for the whole xange of N can therefore be deseribed by: '8(N) /8(5000) =1.3 [1~exp (—8x10-* N)] ” ‘The damage is limited to 1.8 times the damage found after N'=5000. This fac- {or igalao found vsing N=8500 in the 8//¥ relation. The parameter group S/y/N covers a large part ofthe area of interest and will, for sake of simplicity bbe included in the stability formula tobe developed. For N'<1000 and N > 7000 fone ean use eqn, (7) Influence of wave height, wave period and slope angle ‘The extent of damage depends on the slope angle. More atones have to be isplaced for gentlerslopes before the failure criterion, (filter layer visible), reached, asa larger srea ispresent around the water level. Phe lower and upper damage levels, that is the onset of damage and failure, were determined from the investigations and are shown in Table 3. ‘The influence of the wave height, wave period and slope angle is shown in Fig. 7. The dimensionless wave height, H,/4Dyso, is plotted against the surt similarity paramete, ¢,- Kesults are shown for a structure with an imperm able core and after a wave attack of 3000 waves. The damage level, $= plotted for slope angles with cot a-=2, 3, 4 and 6 respectively, The curves are the stability formulse which are derived later in the paper. Figure 7 shows the same trend as that found by Ahrens (1976) for regular229 N= 9000 e6 limparroabe Pig. 7. nfloonce of wave hight pernd and slope angle on stabi waves, Plunging waves are present on the left side ofthe figure for £,<3. Surg- ing waves are shown when £,> 3. Minimum stability is found for the transition from plunging to surging waves, referred to as collapsing waves. For plunging (breaking) waves thesuri similarity parameter, é,, or breaker parameter gives 1 good description of tho influence of slope angle and wave steopness on sta bility, For surging waves on the right side of Fig. 7, different curves are shown for different slope angles. The transition from plunging to surging waves shifts to the right (larger value) for steeper slopes. Influence of permeabiity ‘Three structures have heen tested: A revetment with an impermeable core, clay or sand in nature and conerete in model, tested first. A structure with a permeable core (breakwater) tested second. This structure was much more stable than the impermesble revetment. A homogeneous structure consisting only of armour stones which is an upper boundaty, as far as permeability is ‘concerned, tested third. The impermeable revetment can be regarded asa lower ‘boundary of permeability. The coefficient P was introduced to take account of permeability, described later in the paper. "The results for these three structures are shown in ig. 8. Results are shown for a slope with cot a=2, a damage level of S=5, and after 3000 waves, The influence of the wave period for plunging waves (left-hand side of figure) shows the samo trend for allthree structures, although a more permeable structu more stable, A more permeable structure is also more stable for surging waves (£,>3.5), but the stability increases with larger wave periods. The curves are steeper for larger permeability.ig 8. Inuenc of permeability on stability. ‘This phenomenon can be explained in physical terms by the difference in water motion on the slope. For a slope with an impermeable core the flow is concentrated in the armour layer eausing large forces on the stones during rundown, For a slope with a permeable core the water dissipates into the core and the flow becomes less violent, With longer wave periods (larger ¢,) more ‘water can percolate and flow down through the coro. This reduces the forces and stabilizes the slop. Influence of relative mass density ‘Tests were performed with stones having different mass densities. The light stones (broken bricks) had a mass density of 1950 kg/m° and the heavy stones (basalt) of 3050 kg/m. Normal stones had a mass density of 2620 kg/m’. Results for light, normal and heavy stones are shown in Fig. 9 for the onset of damage, S=2, Both the light and heavy stones are a little more stable than the oa 0.98 206 mas tee g Permeable core Sus £ cota s 2 10 & Fig. 9. Ifoonce of rlativemas dansity on stability.21 ‘normal stones. Within the scatter of the results it can be concluded that the influence of the mass density ean be described by Hy/D,an DERIVATION OF STABILITY FORMULAE “The frat ests f th investigation wore prforine in 1989, The as test. wes finshed in March 198, The results were analyzed inially onthe bass ofthe suumbor of tests performed the test program being changed in acordance vi the resis of this inital analysis ‘The iniil analyses were reported by Van der Meer and Pileresk (984) and Van der Meer (198). All resus have Deen re-analyzed fr the present paper, keeping in mind the intial results and tak ing nt acount the tests performed ater publication of Van dr Meer (1986). ‘The formulae derived in the present report can be regarded ae the definitive result ofthe investigation “The list of variables given in Table 1 ean be shortened using the results described above. The influence ofthe uber of waves is iven by S//N. The influence ofthe relative mass density i given corectly by Hy/4Duay Armour fing, spectrom shay and wave groopiness have praeteally no faftence on the stability (using 7) and can, therfore, be deleted. ‘Thestbiiyof rubble mounll revetments sini breakwaters can then be described by the folowing dimensiones varia: H./ADan coc S/N permeability P AAIl results show a clear difference between plu ‘Therefore two stability formulae have been consider and one for surging weves, ng and surging waves. one forplunging waves Plunging waves ‘The influence of wave period (or steepness) and slope angle can be doseribed by the breaker parameter £,, see Figs. 7-9, These influences can be described by a power function: H,fADseg =a (8) where « and b are curve-fitting coefficients. The coefficiont b was determined by regression analyseson the actual test results and was equal to ~0.54, ~0.51 and ~0.51 for the impermeable core, permeable core and homogeneous stru ture respectively. The faetor b showed no dependency on damage level, slope angle or storm duration. A value of ~0.5 was chosen for b, ‘Phe damage level and number of waves can be described by 8/y/N. This influence of the damage level and the number of waves can also be described by a power function:aa Hi {AD yao X/%1 = ASN)" «) In this case the value of the coefficient, 6, was 0:22, 0.17 and 0.9 for the impermeable core, pormeable core and homogeneous structure, respectively. ‘An average value of 0.2 was chosen. This changes eqn. (9) into: Hi /AD yan Xe = a(S] /N)°* (a0) ta, is only dependent on the permenbility of the structure. ‘The following formulae were derived for the three structures tested: impermeable core: H,/AD go X/E,=441 (S/N) ay 1 /ADase Xv/bx L/AD aso % J Formulae (11) to (13) can be combined i ability coefficient, P, is introduced, permeable core: homogeneous structure: Surging waves ‘The same kind of procedure can be followed for surging waves, although the breaker parameter does not cover the influence of the slope angle. The influ- ence of the wave steepness is described by: HJAD yoo =0 6," ) For the three structures tested, impermeable core, permeable core and homo- ‘gencous structure, the value of the coefficient 6, was 0.1, 0.5 and 0.6, respec- tively. The increasing value of b shows the increasing influence of wave steepness with increasing permeability. This reflected in the steeper curves found on the H,/4Djg0~€,plot, se Fig, 8. The influence ofthe slope angle can ‘be described by: 3: Hy /ADang =0 608 OE, aay Hi JMDysy =2.008 E2® (as) A slove with cot a=2 was tested for the homogeneous structure; the influ tence of slope angle was assumed to be the same for impermeable and permeable core tests, ‘The coefficient in eqns. (14) and (15) was found to be 0.46 and 0.54, respectively. A value of 0.5 was selected for the present study which reaulted in the parameter ./cot a. The influence of damage level and number ‘of waves can then be deseribed by:103 impermeable cove: H,/AD jap —a(S/ JN)" foot a Et as) permeable cove: H,/4Daso =a(S//N)* feat a &.0* an homogeneous structure: H,/AD yon =a(S/y/N)* \/eot a &,°* (18) ‘The coefficient, b, in eqns. (16)—(18) was found to be 0.17, 0.19 and 0.25, respectively. A value of 0.2 was selected for the prosent study. The cootficient, 4, in eqns. (16)~(18) is only dependent on the permeability of the structure. ‘The following formulao were derived for the three structures tested, by curve- fitting ofthe coefficient ain eqns, (16)~(18) and using b=02 impermeable core: H,/AD ao =1.35 (S//N)"? Jootag* ——(19) permeable core: H,/ADsy = 1.07 (S/,/N)®* foot a €,0° (20) homogeneous structure: H,/ADyxy=1.10 (S/\/N)™ Joovaée* (21) ‘The permeability coefficient, P A permeability coefficient, P, was introduced into the stability formulae to take into account the permeability of the structure, see eqns. (22) and (28) ‘This permeability coeificient has no physical meaning, but was introduced to ensure that permeability is taken into account. In eqns. (19)~(21) the power coefficient of &,has a value, dependent on the permeability ofthe structure, of 0.1,0.5 and0.6 respectively. Therefore, Pis defined by P=0.1 forthe imperme- able core, 0.6 for the permeable core, and 0.6 for the homogeneous structure. Now the coefficients 0.1, 0.6 and0.6 in eqns. (19)~(21) can be replaced by this P. Four structures aro shown in Fig. 10 for which three P values, 0.1, 0.5 and 0.6, The other structure has an assumed value of 0. In the absence of other information the selection of the P-value is left to the engineers judgement. Further research will probably provide a more physical definition of P, P can then be calculated for each particular structure ‘The coefficients a, 41,5.3and 5.7 in ons, (11)-(18) and 1.35, 1.07 and 1.10 in eqns. (19)~(21) ean now be described as a function of the permeability coefficient, P. 1¢ must be remembered that the impermeable structure and the homogeneous structure are in fact lower and upper boundaries of permeability. ‘The permeable core is structure lying between the boundaries. ‘The perme: ability coefficient can be included in the formulae a follows H/AD oy XE, =a P“S//N)? for plunging waves (22) and H,[ADyso =a PUS/ IN) Jock @E? for surging waves (23)m4 Poo] fis 1 A Drs ao C= Boe] is toe ae 8Ona0 C232 Fig. 10. Permeability cotfcien sasumptons fr various structures. Curve fitting of a P*in eqns. (22) and (23) to the established P-values and. the eoeftiients in eqns, (21)-U13) and (29)~(21) gives th final formulae. "hoes formulae ar HH/ADynX[%=82 POS /N)%* —forplunging waves (24) sod Me/ADysg=L0P%S//NYS /eotaE! for surging waves (25) ‘The coefficient ~0.19in formula (25) suggests thatthe stability wil decrease ‘rts inereasing permeability. This sm contrast tothe results found in model teste The influence ofthe permeability in the surging waves region is, how ‘ever, described by the factor P~®*é,", a factor which increases in this region with increasing P. Formulae (24) and (25) are shown in Figs. 7 and 8, ‘The transition froma8 es Damage 00 818 1s 20 25 8.0 He/( ADn50x( cot at -999)) HUDSON FORMULA == KO = 4 Fig. 11, Comparison ofthe Hudson Formola with the tet esa, plunging waves, formula (24), to surging waves, formula (25), ean be ealea- lated using: &:=(62P™! fina) ren (26) Depending on slope angle and permeability the tension les between = 2.5 and 4 Comparison the design formulae proposed with tet results ‘The Hudson formula, eqn, (3), only considers the “no-damage” criterion. ‘The Shore Protection Manual (1984), however, gives the influonce ofthe dat ‘age level on the Ke factor. Using the data for rough quarry stone the Hudson formula can be transformed into an expression which takes into account the damage level:206 0 P| % : oo Hte/ Aoasoxv gsxr 21 _ronwuun (24) FoR PLONE waves H,/AD oq = 1.11 (cot @) 47S (27) (for Kp -2 equivalent to 2.6% damage and Hy = 1.27H,). Formula (21) ean dreetl be compared vith th modal tut rnit obtained Inthe present inveatgatin, Figure 11 shows this comparison with the area of Interest forthe designer (between onset of damage an fale. is clea that although the Huson formulas in facta vnsonae average ofthe tet reels, itean only be ed to give ough entimate for aprile case Formule (24) and (25) ae shown on Figs. L2 and 18 together with ll he actual teat results On both gure the vertical axis i the parameter 9) /¥N ‘The dataoranimpataeeble corsa prmeable corsa homogeneous etsuclire, tin forthe data of Thompton aid Shutlr (578) ie detanguised in these figures Intotel more than 660 datapoints have ben plotted in the two igure, Tho stability formals (24) and (25) agree well wth the txt renlls and area ubsianal improvement in comparison with he Hodson formula gues 12 and 19 else provide e hai for describing the accuracy of Poraor © Imps 1 Inpormeadie core Thonpsen !1875) Permeable core | 0.20.3 0.6 1, 0:8 9-7, Hs/adns0 / (P"'? Veota gf) Oe 8 to FORMULA (25) FOR SURGING HAVES Fig 18, Stability formola (25) for surging waves with atta esl, rmulae (24) and (25). The scatter in zeauls is due to the scatter likely to occur in nature and the scatter due to curve fitting. Within 500 datapoints for 15, ‘cases it was found that a lower wave height gave more damage than a 10-20% higher wave height, the other factors being the same. In fact, Fig. 3 shows on of these cases. Two tests in 250 gave a damage factor that was 70% higher than expected. These tests were repeated and then gave the expected damage. ‘This scatter will also be present in nature, ‘The coefficients 6.2 in formula (24) and 1.0in formula (25) give the average ‘curves through the test results. Assuming @ normal distribution the standard deviation of these coefficients can be calculated. his results in a standard deviation of 0.4 (6.5%) for the coefficient 6.2 and 0,08 (8% for the coefficient 1.0. These standard deviations take into account the scatter likely to occur in nature and the scatter due to curve fitting and ean be used in probabilistic design procedures or to establish confidence levels. Formulae (24) and (25) can also be rewritten as reliability fumetions foruse in probabilistic calculations, Tis procedure is described by Van der Meer and Pilarcayk (1987) All results were based on small-scale tests. It was stated before the start of the investigation, however, that large-scale tests should be performed in order to evaluate seale effects on stability. Eleven tests were repeated in the Delft Hydraulies large wave flume, Tests on a slope with cot a =8 and with both a permeable and impermeable core were performed. The linear factor between the small- and large-scale tests amounted to 6.25. The armour layer consisted of 15-55 kg stones. This investigation will be described elsewhere, ‘The conclu- sion, however, was very clear: The large-scale tests showed no influence of scale effects on the stability of the armour layer of rock, in comparison with the small-scale tests Formulae (24) and (25) are based on physical model tests. Encouraging results have also been obtained by Kobayashi and Jacobs (1985) using a math: ‘ematical model for the flow characteristics in the downrush to predict the sta~ bility ofriprap. There is quantitative agreement between these results and the results of Alrens (1975). Ii very difficult, howover, to include different dam- age levels and storm durations in such models and it will be some time before ‘complete mathematical model is available to describe stability of rubble mounds, CONCLUSIONS (1) Practical design formulae have been developed for rubble mound re- vvetments and breskwaters under random wave attack based on more than three hhundred model tests and on the work of Thompson and Shuttler (1975). (2) The stability has been determined in a dimensionless form, using: — the dimensionless wave height parameter: H,/ADyso = the breaker parameter: &, — the slope angle: cot a — the damage as fonction of number of waves: S/N — the permeability of the core: P. (3) Within the conditions tested the following parameters did not influence the stability: = the grading of the armour — the spectrum shape the groupiness of waves. (4) The seatterin the results which is likely to occur in nature and that due to curve fitting can be covered by standard deviations calculated for each for- mula, ‘These standard deviations can be used in probebilistic design. (5) Large-scale tests confirmed the validity of the small-scale tests, No scale20 ‘offects were established on the stability of the armour layer, using ¢ linear scale factor of G25, ACKNOWLEDGEMENTS ‘The research activities presented in this paper were carried out for the Neth- erlands Ministry of Transport and Public Works — Rijkswaterstaat. REFERENCES Arend. 1875, Large Wave Tan Tests of Riprap Stability. CHER C. Techies Memraadum ‘Novi, May 1975. Hoedar, PA, 1960. Stability of rock: breakwaters, Dotorsavhandinga vd Chalmers ‘Techn ska Higsala, Nr. 26, Gctebor, 119. Kobayashi, N and Jacob, BX. 1885. Riprap stability under wave action, ASCE.J. Waterway, Port, Coastal und Ocean Sng 111(8), ‘Losada, MA. ond Ginnerutt, aA, 1878, Tho jolt effet of wave height and period on the stability of rable-moundbeskwatts using Iibarcen's mmber. Costl Eng, 971-06, Pigs, K.W. and Boer, K den, 188, Stability and profile development of core oa thei pplication in cogs engineering. Dol Hydraulic Laboratory, Publ. No 228. ‘Thompson, DIM. and Shater, [M187 Kipeap design for wind wave attack. A laboratory ‘sty in random waves, Wallingford, EX 107 Yan Hijum, Band Placa, KW, 1882, Gravel beaches, Bquiibium profile and longshore ‘wanspor ofcourse mateial unde regular and iregula wave attack, Delt Hyaeaulies Lab ratory, Publ. No. 276 ‘Yonder Neer, JW. 1985. Subilty of rubble mound revetments and brekwatrs under random wave attacks in: Developments in Bteakwates, Proc-ofthe Conon Breahvaters'85‘Thomas ‘Telford, London, pp. 1-164, Yon der Mer, JW and Peay, K.W., 1984. Stbikty of rubble mound slopee under vandom ‘wave attack Proc. th Oestal Bog. Cont, Howston, pp. 2620-2694, Yan der Moor, JW. and Pitty, KW. 1987, Stebibty of breskwatersrnour layers —Deler- ‘miniatie and probabilistic design. Delft Hydsoulies Pubiston No. 278.