NOTES

Janbu, Bjerrum and Kjaernsli's chart reinterpreted

JOHN T. CHRISTIAN
Stotle & Webster-Enginec,ring Corpor-rrfiotl,P.O. Bos2325, Bo.stotl, M A 02107, U.S.A.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by McMaster University on 01/06/15

AND
CARRIER
W. DAVID I11
Bechtel, Inc., P.O. Bos2965, Snu Frar~cisco,CA 94/19, U 3 . A
Received April 22, 1977
Accepted September 22, 1977

The chart developed by Janbu, Bjerrum and Kjaernsli for estimating average elastic settlement
of embedded foundations is examined to determine whether improvenients can be made in the
light of moderncapabilities for stress analysis. The parameterdescribing theeffect ofafinite layer
can be evaluated more accurately from recent results such as those of Giroud. The parameter
describing the effect of embedment was originally taken precisely from Fox's work, but finite
element analyses indicate that Fox's correction substantially overestiniates the effect of embed-
ment. A better estimate can be made by using Burland's correction or, more simply, by ignoring
the effects of embedment altogether.

Le diagramme mis au point par Janbu, Bjerrum et Kjaernsli pourevaluer le tassenient elastique
For personal use only.

moyen des fondations enterries est examine dans le but de determiner si des ameliorations
peuvent &tieapportees en utilisant les nioyens niodernes d'analyse de contraintes. Le parametre
decrivant ['influence d'une couche finie peut &re Cvalue avec plus de precision au nioyen de
resultats recents tels que ceux de Giroud. A I'origine, le palmnetre decrivant I'effet de profondeur
etait determine precisenient a partir du travail de Fox, niais les analyses en elements finis ont
indiqul que la correction de Fox surestimait appreciablement I'effet de profondeur. Une meil-
leure evaluation peut &re faite au moyen de la correction de Burland ou, plus simplement, en
ignorant completement cet effet.
[Traduit par la revue]
Can. Geotech. J . , 15, 123-128(1978)

Introduction erties, can be dealt with, but finite element

Twenty years ago, Janbu, Bjerrum and Kjaernsli methods require fairly expensive computer runs
(1956) published a monograph, in Norwegian, and a considerable effort in preparing input. For
on the application of soil mechanics to several the engineer who needs a reasonable estimate for
problems of foundation engineering. They included his job, the chart is among the very few simple
in it the chart reproduced in Fig. 1 for computing techniques available. It has been reproduced in
elastic settlement of circular or rectangular founda- a textbook on foundation engineering (Bowles
tions embedded at various depths. The settlement 1968) and a compendium of elastic solutions
is to be computed from the equation Lysmer and Duncan 1972); it has even found its
way into the required input for a publicly available
computer program (Almes 1970; Jubenville
1976). Furthermore, and of greater significance,
in which q is the average applied vertical pressure, the chart is found in the notebooks of a large
B is the width of the foundation, E is Young's number of practicing engineers.
modulus, is the settlement, and PO and p1 are Because of its considerable usefulness and wide-
dimensionless parameters obtained from Fig. 1. spread application, an investigation of the sources
The parameter po describes the effect of the depth of the chart, of its accuracy, and of possible im-
of embedment, and p1 describes the effect of the provements is considerably more than an academic
depth of the layer itself. exercise. Several questions have arisen over the
The chart is a very useful tool for estimating the years, including:
settlement under a variety of structures. Since its ( a ) Is the solution intended to be the average
original publication, finite element methods have settlement, the settlement of a corner, or the settle-
developed to the point that more complicated ment of the center of the loaded area, or is it the
geometries, as well as nonlinear stress-strain prop- settlement of a rigid area?
 124 C A N . GEOTECH. J . VOL. IS. 1978

Poisson's ratio of 0.5 only. Comparison of Fig. 1

with thc figure in Fox's paper and with numerical
values obtained from Fox's equations indicates that
,lo is indeed the same as Fox's ratio and that it is

valid for Poisson's ratio of 0.5 only.

The origin of p1 is somewhat more obscure:
there is no reference given in the original publica-
tion (Janbu et al. 1956). However, it appears that
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by McMaster University on 01/06/15

the curves were obtained from the Steinbrenner

approximate method (Terzaghi 1943) with the
settlements averaged over a rectangular area. Eval-
"." uation of the average settlement from the Stein-
0.1 0.2 0.5 1 2 5 10 20 50 100 1000
H/B brenner equation gives values of p1 that are very
1 .o
nearly the same as those in Fig. 1 for H/B greater
0.9
than about 5.
Po ;:", For H/B less than 5, the calculated values of
0.6
p1 are about 75 % of those in Fig. 1, and this sug-
0.5 gests that the factor ( 1 - v"), which is 0.75 for =
11
0,l 0.2 0.5 1 2 5 10 20 50 100 1000
0.5, was inadvertently or deliberately left out of the
D/B
calculation of p1 for Fig. 1. Therefore, a correction
FIG. 1 . Original Janbu, Bjerrum and Kjaernsli ( 1 9 5 6 ) factor of (1 - v2) should be applied for H/B less
chart f o r b,, and /I, with captions translated into English.
than about 5, and this appears to be the reason
( b ) What is the effect of Poisson's ratio V?Most this term occurs in the most widely distributed
For personal use only.

published versions of the chart show the right-hand versions of the chart. However, it must be noted
side of [ l ] multiplied by ( 1 - 1;"). Which form of that:
the equation is correct? ( a ) The ( I - v2) correction should not be
(c) Where did the values of the parameters in applied for H/B greater than about 5, where v =
Fig. 1 come from? 0.5 has already been incorporated into the curves
( d ) Is it possible to improve the charts by in Fig. 1.
using the results of the last 20 years of research ( b ) The factor po was developed for undrained
into settlement analysis and elastic theory? setlements with v = 0.5, so it would be very sur-
(e) Is the basic approach of multiplying two prising if the effects of other values of Poisson's
dimensionless factors valid? ratio could be accounted for by simply multiplying
This paper presents the results of a study of by ( 1 - 119, even though that is a valid procedure
these questions that was originally prompted by for loads on the surface of an elastic half-space.
an attempt to improve the charts. In the course of
their intermittent research, the authors have found Improvements in the Parameters
that the chart is still almost the only simple, direct The British gcotechnical community has recog-
aid to estimating settlement when both embedment nized for some years that Fox's correction for em-
and a finite depth of soil are thought to be im- bedment may overstate the case. Burland (1970)
portant. It was therefore somewhat dismaying to has proposed the revised values of po shown in
discover that the procedure is rather seriously in Fig. 2. These are based on the results of finite
error. element studies conducted for circular loaded areas.
They show substantially less effect of embedment,
The Sources of the Parameters especially for deeply embedded foundations.
The original monograph gives a clear reference Since 1956, a number of people have evaluated
for the source of p,,. It derives from Fox's (1948) the settlement of a rectangular loaded area on the
computation of the ratio between the average settle- surface of an elastic layer by means of digital com-
ment of a vertically loaded horizontal area at some puters, avoiding the Steinbrenner approximations.
depth within a semi-infinite elastic half space and In particular, Giroud (1972) presents values for
the average settlement of the same loaded area if the average settlement of a flexible loaded area,
it were on the surface of the same half space. Thus, and these have been plotted in Fig. 3. The curves
the parameter is the ratio of average settlements of are similar in shape to those for p1 in Fig. 1. They
flexible areas. Fox evaluated the parameter for a have lower ordinates for small values of H/B, as
 ( V = 0.5) by means of the finite element program

DAIS, which uses the elastic incompressible for-

mulation of Christian and Watt (1972). Values
of D / B were 0.3, 1 and 3, and values of H / B
were also 0.3, 1 and 3. The finite element mesh
contained 146 elements, and the various depth
ratios were achieved by expanding or contracting
the model in the appropriate directions. Rectangu-
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by McMaster University on 01/06/15

lar loaded areas with L / B of 1, 2 and 5 were

studied by means of a three-dimensional finite
element program using eight-noded isoparametric
elements. Again, D / B was 0.3, 1 and 3, but H / B
was 1, 2 and 10. The finite element mesh con-
tained 870 elements. Because this program cannot
handle incompressible materials, Poisson's ratio
FIG.2. Effect of embedment for a circular loaded area was set to 0.49.
in a semi-infinite half space, after Fox (1948) and Burland The results of finite element calculations are sub-
(1970). ject to errors, because of both rounding in the
would be expected because the effects of a Poisson's numerical calculations and discretization of the
ratio of 0.5 are already included in the curves of continuum into finite elements. However, these are
Fig. 3. It should be noted that the curves cross for certainly the most accurate procedures available to
H / B between 0.5 and 1. compute deformations of an elastic body when no
For personal use only.

The plots of Figs. 2 and 3 are up-to-date ver- analytical solution is available. For purposes of this
sions of pO and p l . It still remains to be determined study, the average settlements from the finite ele-
whether the approach of multiplying pOand p1 gives ment calculations are assumed to be valid, with the
reasonable answers. note that for large L/B and H / B , the results for
the rectangular case can be expected to be less ac-
Comparison of Results from Charts and from curate than for the other conditions.
Finite Element Calculations The accuracv of the various forms of the charts
T o evaluate the accuracy of the procedure with is compared through ratios of the average settle-
the original or improved values of pO and p l , a ments to those computed by finite elements. Five
series of finite element analyses was performed. The ratios were computed and are plotted to the same
average settlement of a circular loaded area was scale in Figs. 4 and 5: (1 ) The ratio of the average
calculated for an incompressible elastic material settlement from the original Janbu, Bjerrum and

3 . Average settlement of flexible loaded area on the sur,face of qn elastic layer, after G ~ r o u d(1972).
 CAN. GEOTECH. J . VOL. 15, 1978
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by McMaster University on 01/06/15
For personal use only.

FIG. 4. Settlement from Janbu, Bjerrum and Kjaernsli (PJIX) compared with settlement from finite elements
( ~ F I : )(.a ) Simple ratio of settlements. ( b ) Including correction factor of 0.75.

1.2 I I I I I Ill
Circle .
PG-B/P~~ Rectangles - L/B = I -------
10 L/B = 2
L/B = 5

FIG.5. Settlement from Giroud ( P C ) ,Giroud plus Fox ( ~ G - F )and

, Giroud plus Burland ( p ~ i compared
) with
settlement from finite element ( ~ F x )(.a ) Giroud, completely ignoring embedment. ( b ) Giroud plus Fox correc-
tion for embedment. ( c ) Giroud plus Burland correction for embedment.

Kjaernsli charts, p.rns, to that from finite elements, PG, to pr.E (Fig. 5a). (4) The ratio of the settle-
pIYE (Fig.4a). ( 2 ) The ratio p . l ~ I ; / p ~ .multiplied
I< by ment evaluated from Giroud's results modified for
0.75 to account for Poisson's ratio as is done embedment by Fox's correction, PGLI.', to P F E (Fig.
by Almes (1970), Bowles (1968), Jubenville 5b). ( 5 ) The ratio of the settlement evaluated
(1976) and Lysmer and Duncan (1972) (Fig. from Giroud's results modified for embedment by
4b). ( 3 ) The ratio of the settlement evaluated Burland's correction (for v = 0.49), p ~ - , < , to P F E
from Giroud's results alone, ignoring embedment, (Fig. 5c).
 Figure 4 shows that the original charts do not to contribute very significantly to the settlement
give an accurate estimate of settlement. This is behavior.
true regardless of whether the factor (1 - 2) is It is to be expected from the general form of
used. Errors can exceed 100% and commonly solutions for embedded loads and for loads on finite
exceed 20%. Use of a better estimate of p1 together layers that the solution for a case involving both
with Fox's values for po (Fig. 5b) improves the effects cannot be obtained theoretically by simply
accuracy, but errors still exceed 20% in many multiplying dimensionless constants from the solu-
cases. Further, the errors are unconservative in the tions for each effect individually. The computer
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by McMaster University on 01/06/15

sense that the settlements are underestimated. results presented here indicate that such an ap-
Completely ignoring embedment, as is shown in proach does not yield adequate numerical approx-
Fig. 5a, results in estimates that are as accurate as imations, either. More accurate results can be ob-
those obtained by using Fox's correction, and the tained from empirical factors such as Burland's or
errors are on the conservative side. by ignoring the effects of embedment.
The best results are obtained when Burland's
correction factor is applied to Giroud's values Conclusions
(Fig. 5c). All errors are less than 2 0 % ; most are An improved version of the Janbu, Bjerrum and
less than 10%. This is not surprising as Burland's Kjaernsli chart, shown in Fig. 6, is composed of
factors are back-calculated from finite element Giroud's results for the effects of depth and
results. Burland's results for the effects of embedment.
It should be noted that Burland's factor was Such a chart can be used to estimate settlements of
computed for circular loads only. Figure 5c shows structures founded on soils that can be assumed to
that the accuracy of results obtained using his factor be elastic and incompressible. Comparison of test
is independent of the shape of the loaded area calculations with the new chart and by finite ele-
For personal use only.

within the range of shapes used in these calcula- ments for values of H / B ranging between 0.3 and
tions. 10, L / B between 1 and 5, and D / B between 0.3
The authors conclude that when embedment and 3 gave errors usually less than 10% and always
effects are to be considered, Burland's correction less than 2 0 % .
factor should be used for po, and Giroud's results The calculations also demonstrated that com-
for pl. Ignoring embedment altogether is nearly as pletely ignoring the effects of embedment (i.e. set-
good a procedure and may be the best approach ting po to unity) gives reasonably satisfactory
when other effects, such as inhomogeneity and non- results in most cases.
linearity of materials, are to be considered. The
material above the foundation level does not seem Acknowledgements
The axially symmetric finite element computer
runs were performed at Stone & Webster Engineer-
ing Corporation. The three-dimensional finite ele-
ment computer analyses were sponsored by Bechtel,
Inc., and were performed with the Structural Anal-
ysis Program (SAP).
ALMES,R. G. 1970. SETTLE. Computer program documenta-
tion, E. D'Appolonia Consulting Engineers, Inc., Pittsburgh,
PA.
BOWLES, J . E. 1968. Foundation analysis and design.
McGraw-Hill, New York, NY. pp. 91-92.
BURLAND, J . B. 1970. Discussion, Session A. Proceedings,
Conference on 111 Situ Investigations in Soils and Rocks.
British Geotechnical Society, London, England. pp. 61-62,
CHRISTIAN, J . T., and WATT, B. J . 1972. Undrained visco-
elastic analysis of soil deformations. Proceedings, Sym-
posium on Applications of the Finite Element Method in
Geotechnical Engineering, Vicksburg, MI., May. pp.
533-577.
Fox, E. N. 1948. The mean elastic settlement of a uniformly
loaded area at adepth below the ground surface. Proceedings,
Second International Conference on Soil Mechanics and
Foundation Engineering, Rotterdam, Vol. I. pp. 129-132.
FIG.6. Improved chart. GIROU D,J.-P. 1972. Settlement of rectangular foundation on soil
 I28 CAN. CEOTECH. J . VOL. IS. 1978

layer. ASCE Journal of the Soil Mechanics and Fo~~ndations of Cololxio Computing Center Report 76-10. Bo~~lcler, CO.
Division, 98 (SMI), pp. 149-154. LYSMER, J., and DUNCAN, J . M. 1972. Stresses and deflections
J A N B UN.,
, BJERRUM, L., i~ndKJAERNSLI, B. 1956. Veiledning in foundations and pavements. Department ofcivil Engineer:
ved losning av f~~ndamenteringsoppg>~ver. Norwegian ing, University of California, Berkeley, CA. Fifth Edition,
Geotechnicnl institute Publication 16, Oslo. pp. 30-32. p. 65.
JUBENVILLE, D. M. 1976. SETTLE-11, version I-A, acomputer T E K Z A G HKARL.
I, 1943. Theoretical soil mechanics. John
pi:ogram to calculate settlements, ~tser'smanual. Geotechni- Wiley & Sons, New York, NY. pp. 423-427.
cal Engineering Software Activity Report D-76-6, University
. .
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by McMaster University on 01/06/15

A pictorial case history of lateral rock creep in an open cut

into the Niagara Escarpment rocks at Hamilton, Ontario

AND
For personal use only.

Received September 15, 1977

Accepted September 28, 1977

Excavation of an open trench to a total depth of 10- 13 m including a 5.5 m into the
Er;~mosaMernber of the Lockport dolomite formation resulted in an estimated 5- 10 cm of lateral
'
rock movement towards the excavation with slippage clearly defined at bedding planes and joints
as illustrated in the accompanyingphotographs. The movements observed are the visual manifes-
tation of release of high locked-in lateral stresses characteristic of rocks of the Niagara escarp-
ment. If movement is not allowed, very high stresses are ultimately tlmsfen-ed to any rock
encased structures.

L'excavation d'une tranchee B ciel ouvert realisee jusqu'i une profondeur de 10 B 13 m,

incluant une penetration de 5.5 m dans le membreEramosa de la formation dolomitique de
Lockport, a produit un mouvement lateral de la face rocheuse de 5 B I0 cm vers la fotlille avec des
glissements bien definis le long de plans de lits et de joints tels qu'ill~~stres
sur les photographies
jqintes. Les mouvements observis sont la manifestation visible du relichement des fortes
contraintes laterales bloquees dans la couche rocheuse et q ~sont ~ i caracteristiques de I'escarpe-
ment du Niagara. Lorsqtt'auc~in mouvement n'est possible, de trts fortes contraintes sont
transmises tlltimement a toute strLlcture enfouie dans le roc.
'
[Traduit par la revue]
Can. Geotech. J.', 15, 128-133 (1978)

Recent engineering studies by Palmer and Lo rock before construction, the presence or absence
(1976) have shown that horizontal stresses as high of compressible construction layers installed to
as 14500 kPa (2100 psi) exist in the Silurian age, absorb the inward deformation, the rigidity of the
cap rocks of the Niagara escarpment. Excavation structure and ultimately the magnitude of the in
of tunnels, shafts and trenches into these rocks situ rock stresses. Cascs of rock squeeze relevant
results in stress release and lateral movement, to the Niagara area have been discussed by Lo
which decreases in rate with time but may proceed et al. (1975), Lee and Lo .(1976), Palmer and
in some cases for decades (Lee and Lo 1976). Thc Lo ( l 9 7 6 ) , Bowen et al. (1976) and Franklin
amount of damage done to rock encased structures and Hungr (1977).
by the gradual buildup of rock stresses depends on The purpose of this technical note is tb describe,
the amount of time and movement allowed in the with the aid of several photographs, the visual cx-