Sie sind auf Seite 1von 24

Repair of Historic Timber Structures

David Yeomans
ISBN 978-0-7277-6438-6
ICE Publishing: All rights reserved

Chapter 8
Decay and repair
Classification of repairs
It is useful to begin with some classification of repairs to identify the issues that need to be
considered in design. For our purposes here, repairs can be classified as follows.
(a) Those where timber is replaced with timber. In such cases, the repair might be effected
with carpentry alone. Today, ‘carpentry’ includes the possibility of some kind of
mechanical fastening.
(b) Those where timber is supplemented with some other material, either a steel plate,
carbon fibre or some other reinforcing material. These are usually for the simple
strengthening of a member that has inadequate load-carrying capacity but may also be
used to connect new timber to old where part of a member has to be replaced.
(c) Those where steel or some other material is used to replace the timber. Such repairs are
most often used where there are decayed ends of floor beams or tie beams and principal
rafters of roofs. They have the advantage that less historic material is lost but can only
be used if appearance of the repair is not of concern.
(d) The use of supplementary structures that act either independently of the original,
simply providing it with support, or that act in consort with the original structure,
although altering its structural behaviour. Such supplementary structures could be of
either timber or steel. The timber structure inserted into the apse roof at Westminster
Abbey by Hawksmoor is a supplementary structure. Supplementary queen post trusses
were inserted into the choir roof at Chichester during the eighteenth century and
bracing to try to restrain racking has been inserted in the west end of Exeter Cathedral

Design codes
Present-day design codes are intended for the design of modern structures and not for the repair
of historic structures. Here, we may usefully be reminded of our medical analogy. When I visit
the doctor, she does not try to return me to the state of that active adult in his twenties nor even
to the active middle-aged person. Instead, she has to cope with someone who still wants to be
able to garden and walk to the station without difficulty. To do so my body requires a little more
considered maintenance and perhaps more considered maintenance is something that an
historic structure might require. Carrying out repairs might not be the complete answer and
some regime of continuing care might also be appropriate.
The parallels are that one should not be trying to bring a historic structure up to modern standards,
that is, to conform to present-day design codes. The task is to give it a helping hand, and any
intervention might involve the assumption of regular maintenance. Because codes are developed
for today’s designs, it is essential to take a wider view than simply relying on them completely.
Their limitations should be recognised and recourse might have to be made to other sources.

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

Some may prefer to see traditional joints used in repair, even if they have to be supplemented
by mechanical fasteners. However, in selecting the type of joints used, one needs to consider
the method of assembly and possibly the need for some partial dismantling of the structure.
This is the kind of judgement that requires some discussion between the engineer and the
others involved in the process.

Mechanical fasteners
Much of this chapter will deal with repairs made with mechanical fasteners. Here we should take
note of the opening words in the chapter on joints in the Institution of Structural Engineers
manual for Eurocode 5 (IStructE, 2007). This says, ‘The Task Group took the view that the EC5
design rules for timber connectors were unsafe, so a method based on BS5268-2 for obtaining
safe design resistances for these devices is given instead.’ That is the approach adopted here.

Mechanical fasteners are generally nails, bolts and screws, the last two being the more
common in conservation work. It is also possible to use split-ring connectors or shear plates
but these are probably of limited use in repair work (see later). Screws are generally more
convenient than bolts because access is only required on one side of the joint and if the fastener
is to be countersunk and pelleted there is less work to be done. There are also many situations
in which access is only possible from one side of the joint. Therefore, this chapter will con-
centrate on the use of screws.

Tables are provided in BS5268 (BSI, 2002) for the allowable loads on fasteners of different
diameters and in different thicknesses of timber but these, being designed for modern con-
struction, make assumptions about timber thicknesses that are not always appropriate for early
carpentry. Moreover, some engineers seem to have been reluctant to use the sizes of coach
screws larger than those covered by BS5268, that is, only up to 10 mm diameter, in spite of the
advantages that they give over bolts. The solution to these difficulties is to use the formulae
provided in Annex G of the code for both bolts and coach screws to determine the safe loads.
These derive from Johansen’s analysis of the possible modes of failure, either through crushing
of timber in compression under the load of the fastener (called embedment) or crushing of the
timber accompanied by a yield failure of the bolt or screw (Figure 8.1). There is, however,

Figure 8.1 Failure mode of bolts and screws used as the basis for Johansen’s equations

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

a difficulty in using these formulae for assessing the capacity of existing joints because one
does not know the yield capacity of the historic bolts.

Apart from the properties of the fastener itself, the density of the timber has an effect on joint
capacity. It will presumably be possible to identify the softwood species used in the original
structure, and certainly those specified for repair, and so apply the appropriate density when
using the formulae. Lavers (1967) gives densities for Quercus rubra (oak) as 705 kg/m3 when
dry. BS5268-2: 2002 (BSI, 2002) gives figures of between 680 and 713 kg/m3, depending on
the grade. However, only the species and not the grade of any timber, whether hardwood or
softwood, is significant as the carpenter should be making a repair in clear, that is, knot-free
timber, whether for the repair timber or in the original. It is not acceptable to make a repair in
the vicinity of a knot.

The load capacity of traditional pegged joints has been discussed in Chapter 3 in relation to the
possible performance of timber frames. Of course, in early timber frames these pegged joints
rarely carried tensile forces but it is possible to assemble joints for repairs using pegs as
fasteners. They might, for example, be used in either scarf joints or bridle joints (see later)
instead of bolts or screws. If the joint is visible, they have the advantage of presenting an
acceptable appearance without the necessary countersinking and pelleting associated with the
heads of screws. The pegs can, of course, be used to pull joints tight just as they were in the
past, especially in bridle joints. If the shear load on the peg is low, this is a sensible type of
fastener. Pegs should be cut flush with the timber and not left projecting. The result of making
‘traditional’ repairs with pegs left projecting is simply ugly. No self-respecting carpenter
would have left a joint looking like that.

Tables within the code provide allowable loads on bolts for the full range of bolt diameters that
the conservator is likely to use. Although they only deal with timber up to 147 mm thick, while
some historic timbers are much larger than this, this is not a problem because the allowable
loads remain constant beyond a certain thickness, as the failure load is dominated by bending
of the bolt. For softwoods, the allowable loads perpendicular to the grain are less than those
parallel to the grain but the same is not true of oak, where the values are the same.

It has been known for contractors to want to use threaded bars with nuts put on both ends
instead of bolts because the bar can be simply cut to the length required. This is not satisfactory
for shear loads because there will be some slippage of the joint as the load is applied and the
threads embed themselves into the timber. Also, the bending strength of the stud is signifi-
cantly less than that of the bolt. A threaded bar can be used where the fastener is to act in

Both the spacing, edge distances and end distances of bolts are not to be closer than those
shown in Figure 8.2 to avoid splitting of the timber.

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

Figure 8.2 End and edge distances of bolts and screws

4d unloaded end 4d
7d loaded end




4d B




Screw technology has been changing in recent years, with the introduction of such devices as
self-tapping screws for timber (see later). There is also a distinction between modern screws, in
which the thread is larger than the shank, and earlier screws, in which the shank was the outside
diameter of the screw with the thread cut into it. Manufacturers provide data for the capacity of
their screws, while coach screws (lag screws in the USA) may be used where larger diameter
fasteners are required. It is for the latter that Johansen’s equations might be used. Note that the
load capacity of a group of fasteners needs to be reduced for close spacing, to avoid splitting of
the timbers. Generally, the reduction applies if the screws are closer than ten diameters parallel
to the grain or three diameters perpendicular to the grain (BS5268, clause 6.5.3 (BSI, 2002)).
However, screws of 8 mm diameter or larger may be treated as bolts, with the spacing as close
as four diameters parallel to the grain, as in Figure 8.2, before any reduction has to be applied.

Like bolts, coach screws are used with a washer under the head and require a larger pilot
hole than ordinary screws. While BS5268 clause 6.5.1 (BSI, 2002) says that screws should

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

have a pilot hole half the shank diameter, this surely only applies to ordinary wood screws.
The Wood Handbook (Forest Products Laboratory, 1987, Section 7.11) recommends that,
for lag screws, ‘The lead hole for the shank should be the same diameter as the shank. The
lead hole for the threaded part varies with the density of the wood.’ For denser softwoods,
which are those likely to be found for structural use in historic buildings, it recommends a
pilot hole for the screw 60–75% of the shank diameter; for hardwoods, it states that it
should be 65–85% of the shank diameter. The Wood Handbook further recommends the
use of lubricants for inserting the screws and slightly larger lead holes for longer screws.
For some purposes, these screws are used in withdrawal where either washers or steel
plates are used under the head.

Experimental work carried out in the USA (Forest Products Laboratory, 1987) showed the load
capacity of screws in both withdrawal and shear to be a function of the density of the timber. For
both ordinary screws and lag screws, the lateral load that could be carried was shown to be given
by F = Kd2, where K is dependent on the density of the timber. These values assumed the screw
length in the foundation timber to be 7·5 diameters, with the shank length in the attached timber
to be half of that and the formula applied to both ordinary screws and lag screws. The values
given for ordinary screws in the 1984 edition of BS5268 (BSI, 1984) used this formula and the
values for K were given by Baird and Ozelton (1984), who noted that no information was
provided for coach screws but suggested the use of the same formula. Note that the allowable
load on a screw is reduced if either the shank or screw length is less than the standard length.

Only one set of figures is provided for the lateral load on screws in BS5268 (BSI, 1984) so the
assumption must be that allowable loads on ordinary screws are independent of the direction of
loading. This is not an acceptable assumption for coach screws used in softwood, where there
are smaller allowable loads for bolts perpendicular to the grain. Therefore, it seems prudent to
apply the same reduction factor to both coach screws and bolts. In the absence of any other
guidance, the author has assumed that the formulae given apply only to loads parallel to the
grain and has applied a reduction factor for load perpendicular to the grain. This has been
derived by taking the ratio of allowable loads on bolts of the same diameter in timbers of
thickness 3·5× the screw diameter. This is a little more generous than the reduction factors
suggested in Wood Handbook (Forest Products Laboratory, 1987).

Based on this assumption, Table 8.1 provides loads for coach screws between 8 and 12 mm in
both oak and in C24 softwood both perpendicular and parallel to the grain based on this approach.
For all other angles, Hankinson’s formula (see the appendix) should be applied. The explanation

Table 8.1 Allowable loads for coach screws: N

Screw diameter

8 mm 10 mm 12 mm
Oak 1849 2285 3258
C24 parallel to grain 1391 1741 2454
C24 perpendicular to grain 1280 1515 2135

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

Figure 8.3 Screw lengths for full loads

1/3 3·5× diameter

2/3 7× diameter

for the difference between oak and softwood is that a principal factor affecting the load that either a
bolt or screw will carry is the embedment strength. In softwood, this is lower when the material is
loaded perpendicular to the grain but in oak is very similar whatever the direction of loading.

It is important to control the workmanship in making joints with these fasteners. Load carrying
depends on the minimum head and point lengths being achieved (see Figure 8.3). The screw
must not be countersunk into the timber to a depth that would compromise the former and must
be long enough to achieve the latter. Any reduction in these two distances results in a corre-
sponding reduction in the allowable load.

Screws in tension
In the design of some joints, such as scarf joints carrying bending moments, screws act in
tension to hold the two pieces of timber together. Thus, we are relying on the purchase of the
screw in one piece of timber and the compression under the head of the screw, and its asso-
ciated washer, in the other. The withdrawal load of the screw is expressed in terms of the load
per millimetre of penetration of the screw into the timber and is dependent on the diameter of
the screw and the density of timber.

If a standard washer is used that is twice the diameter of the screw shank, then the allowable
load can be assessed by multiplying the bearing area by the allowable stress in compression
perpendicular to the grain. In practice, one might assume a slightly higher load because the
timber here could be regarded much as a short bearing (see BS5268 clause 2.10.2 (BSI, 2002))
but there has been no experimental work to verify this. As large loads can be generated by a
suitably long screw, it may often be found convenient to use a small steel plate rather than a
washer. The thickness of the plate will be determined by bending stresses.

Self-tapping screws
There has been considerable development in the design and understanding of self-tapping
screws in recent years. These, as the name indicates, can be driven in without the need for a

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

pilot hole. This is normally in softwoods, although there are some screws designed for use in
hardwoods. It is also possible to use a pilot hole and that might be an advantage in ensuring
accuracy when the screw is to be driven at an angle, which is difficult if attempted free-handed.
It might also be possible to reduce the spacing with pre-drilling. Some self-tapping screws
have a plain shank (and possibly a shank cutter) above the threaded portion but the most useful
are likely to be those that are fully threaded. Because of the wide variety of screws available it
will be necessary to refer to the manufacturers’ information.

The kinds of repair or reinforcement for which these may be useful is in reinforcing a notched joint
or attaching replacement timber, that is, a prosthetic, at a beam end. For the former, there will be
shear stress at the notch as well as tension across the grain (see Figure 8.4). Moreover, the shear
stress will be greatest close to the root of the notch. For this reason, it is considered preferable to
drive the screw at 45° to the grain. This enables the screw to be closer to the root of the notch than if
it were driven vertically, and at that angle the screw is better able to carry the shear force.

The tensile force perpendicular to the grain can be approximated by integrating the shear stress
in the shaded area of Figure 8.4(b). However, because the tensile stresses are increased as a
result of the eccentricity between the support and the inner edge of the notch, for x ≤ h′/3, the
tensile force perpendicular to the grain should be increased by a factor of 1·3. This design
tensile force is then given by
F = 13V 3ð1 − αÞ2 ð1 − αÞ3

where V is the design shear force.

Figure 8.4(c) shows the screw inserted at 45°, which minimises the distance a between the screw
and the root of the notch. As only one row of screws will be effective for large shear forces, it may
be necessary to place several using the spacings indicated in Figure 8.5 and Table 8.2.

a1 and a2 may be reduced to 2·5d if a1a2 ≥ 25d2

For further information on self-tapping screws, see Harte and Dietsch (2015).

Figure 8.4 Self-tapping screws to reinforce a notched joint

h’ lb


(a) (b) (c)

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

Figure 8.5 Spacing and edge and end distances for self-tapping screws




Table 8.2 Typical minimum spacings and distances for axially loaded self-tapping screws where
d = screw diameter

a1 a2 a3 a4

5d 5d 5d 4d

Split-ring and shear-plate connectors

These are used in association with bolts and the grooves for the ring or plate are cut with a
special cutter once the hole for the bolt has been drilled. They are normally used in the
workshop, where the grooves can be cut on a bench-mounted drill. However, with care, it is
possible for them to be cut on site. While split-ring connectors are normally used in association
with bolts, they may also be used with coach screws. If one compares the allowable load on
split rings with that for coach screws, it will be found that a group of coach screws can be
placed within the area required for a split-ring connector that will resist a similar load.
Therefore, there seems little point in using a split ring.

Shear-plate connectors that are used in association with bolts and steel plates carry a larger load
than a split ring of the same size but have the same requirements for end and edge distances.

Moisture content and fasteners

The basic load on bolted joints applies to timbers used at an average moisture content below
20%. For timbers that have a moisture content above this, a reduction factor of 0·7 should be
applied to the basic load to allow for the weakness of green timber. (This factor is applied to the
allowable loads for all fasteners.) If a bolted joint is made while the timber has a moisture
content above 20% but the timber then dries in service to an average moisture content below
this figure, a reduction factor of 0·4 should be applied. This is a substantial reduction, to be
avoided if possible, and one can see that it is greatly to the advantage of the structural design to
be able to obtain dry timber. (Of course, there is the complication in repairs that one-half of the
fastener may be in green timber, and the other half in dry timber.) The logic of this reduction
factor is that the shrinkage reduces the size of any hole drilled in the timber as well as its overall

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

dimensions. While the size of the hole will reduce, that of the bolt or coach screw within it will
not, resulting in outward forces on the circumference of the hole and possible splitting of the
timber (see BS5268 clauses 6.6.6 (Permissible load for a joint) and 1.6.4 (Service classes)
(BSI, 2002)).

Although this rule only applies to bolts, with no figure given for screws used in the same
circumstances, coach screws should be treated as if they were bolts for this purpose, especially
as it is often the plain shank of the screw that will be in the green timber. In the absence of any
firm guidance in the code of practice on this matter, it would seem sensible to simply allow for
the shrinkage in making the hole, that is, to drill the hole for the bolt or coach screw slightly
oversize. With a shrinkage of 5% from green to dry, this means a maximum of 0·5 mm for a
10 mm fastener. This is contrary to the normal recommendation for fasteners in dry timber,
where the hole diameter should be as close as possible to that of the fastener. Conversely, if the
timber does split, the load the screw can carry presumably falls to zero.

Groups of fasteners in joints

In some cases, large moments may have to be transferred from one timber to another, larger
than is possible with a single pair of fasteners. For example, Figure 8.6 shows a drawing of a
joint required in a main cross beam of a timber frame where the end of the repair had to carry a
main post because of the small quantity of original material remaining at this point. However,
this load would eventually be transmitted into the original timber and carried as a bending
moment with several fasteners involved in the joint. If it had been a steel connection, one
would have been happy to carry out an elastic analysis of the joint, distributing the fastener
forces accordingly. However, with the extent of movement associated with timber fasteners at

Figure 8.6 Group of fasteners used in a repair of a main beam

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

high levels of load, one can hardly expect a timber joint to behave in a conveniently elastic
way. In the event, it was assumed that all fasteners would be loaded to their maximum in
calculating the resistance of the joint.

Adhesive bonded reinforcement

Epoxy-resin-based adhesives and mortars were originally introduced to Britain with no sup-
porting research to justify their use and a poor record of quality control in their application.
Applications involved the bonding of steel plates into beams, the bonding of steel or fibre-
reinforced plastic rods into holes drilled into beam ends and the use of adhesive bonded
mortars to form a prosthetic to replace decayed beam ends. While some applications have
proved successful, poor quality control in some cases resulted in poor performance, sometimes
causing more harm than good. Occasionally, there was failure to limit the run of the adhesive to
where it was wanted. However, there has been some improvement in recent years, although
work still remains to be done. And, of course, we cannot know the long-term performance of a
material that has been used for only a relatively short time. It has only been about 60 years
since these materials were introduced.

Several materials have been used for the reinforcement, including steel plates, threaded rods,
used rather like reinforcing bars, and fibre-reinforced plastic. Threaded rods have the
advantage that there is a mechanical link between the adhesive and the reinforcement.
Although steel or fibre-reinforced plastic rods, with glass as the basic material, were used
initially, other materials have since been introduced. Carbon fibres come in two forms,
depending on the method of manufacture: high strength and high modulus of elasticity.
Aramid (aromatic polyamide) and basalt fibres are also available. Typical relative properties of
these are shown in Table 8.3 (Harte and Dietsch, 2015), which shows that carbon has the best
performance, in terms of both stiffness and strength.

The performance of any repair or reinforcement using an adhesive depends on three com-
ponents: the timber, the adhesive and the reinforcing material. For the repair to work, the
adhesive must bond with the timber and for that the surface must be free of dust. Thus, any
drillings or slots cut in the timber must be cleaned out before the adhesive is applied.

Table 8.3 Typical properties of materials used for reinforcing

Material Modulus of elasticity: GPa Tensile strength: MPa

Glass 70–80 2000–4800

Aramid 62–180 860–3450
Basalt 82–110 860–3450
Carbon (high modulus of elasticity) 390–760 2400–3400
Carbon (high strength) 240–280 4100–5100

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

Although other adhesives have shown satisfactory bonding performance if used in controlled
environments, two-part cold-cure epoxy adhesives are the most suitable for on-site use. These are
thixotropic, have good gap-filling properties and low curing shrinkage. However, epoxies have a
poor performance when subject to repeated wetting and drying and so are not suitable for use in
exposed conditions, where adhesives such as polyurethane might be used instead. Moreover, the
moisture content of the timber at the time of bonding should be below 20–22%.

The two-part resin comprises the active ingredient plus a hardener, which must be mixed in the
correct proportions. Commercial formulations also include additives, such as fillers, which
reduce the quantity required of the expensive epoxy, and thickeners or diluting agents to
change the viscosity of the mixture. These do have an effect on the mechanical characteristics
of the cured joint and the suppliers of these adhesive products should be consulted on the
appropriate mixture for any particular situation. Phenol-resorcinol-formaldehyde resins, which
are used in factory conditions are not suitable for on-site applications because of their limited
gap-filling properties and their requirement for controlled pressure during curing.

Reinforcement can be used to increase the capacity of a beam if an increased floor load is
required, to bridge across a fracture in a beam or to attach a prosthetic, either to replace a
decayed beam end or post foot. If attaching a prosthetic, the reinforcement can be in the form
of rods let into drilled holes in the original timber, the advantage being that there is no dis-
turbance of the face of the original sound timber. However, the difficulty is in ensuring that the
holes are properly cleaned out and that the reinforcing is coaxial with the hole so that there is a
uniform thickness of adhesive between the reinforcing rod and the timber.

The alternative, and the method used for other situations, is to chase the timber and let the
reinforcement into that. It is then easier to ensure proper placement of both reinforcement and
adhesive, and if the chase is made deeper than necessary for the reinforcement, a timber strip
can be placed in the top of the chase to restore the original surface. This is not only visually
desirable but, because epoxy resins perform poorly in fire, the insulating properties of the
timber strip provide a degree of protection. This implies that the depth of the strip should be
chosen in relation to the charring rate to provide the required fire rating.

Timber replacements
Where there has been replacement of part of a timber section that is under load, there is the
question of how the loads will now be distributed between the original timber and the new
timber. Can, for example, the bending stresses be distributed in the same way as they would be
in a solid section or does either the original timber or the repair material carry the load pref-
erentially? Presumably, this depends on the fastenings between the new material and the
original or on how the load is applied to the member.

A patch at a mortice
This is a fairly common repair in plates or rails for damage as a result of water becoming
trapped within the joint or of weathering of the surface. The problem is often exacerbated by

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

Figure 8.7 Face patches in a beam



unsuitable materials being used for the infill between the studs, and twentieth-century repair
work has often used cement renders on the outer walls, resulting in more water finding its way
into these joints. Naturally, the repair may involve replacement of the end of the stud or a patch
in the timber containing the mortice where a stud tenons into a horizontal member. The
function of the joint is to transmit compressive forces across the shoulders of both the tenon
and the mortice. If, as is not uncommon, the outer surface has weathered while the inner
surface remains sound, a patch will only be required for the external timber (Figure 8.7(a)),
fixed with adhesive or pelleted screws.

Repair work carried out at the Weald and Downland Museum has involved letting replacement
timber into the original and fastening it in place with epoxy resin adhesive (Figure 8.7(b)). The
intention was to preserve as much of the original surface as possible so that this remains as a
thin ‘veneer’ over the patch. This was possible because the structures were dismantled and the
repair made in the workshop. Whether there will be circumstances in which this technique can
be adapted to in situ repair work remains to be seen. Nevertheless there are similar structural

In ideal circumstances, the load in the stud will be transmitted evenly across the shoulders of the
joint but we cannot assume this to be the case. A bare-faced tenon, as shown in Figure 8.7(a), will
clearly bear on one side of the mortice only but this might also be true of a normal tenon and is
especially likely after repair. The fresh surface of the repair timber on one side and the historic
timber on the other may well favour the former in transmitting the load and replacement of the
joining timber will make this even more likely. Therefore, the fasteners between the repair and
the original material should be designed to carry the full load.

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

Figure 8.8 Braces on patched posts to show two different cases: (a) brace in the same plane as the
repair joint; (b) brace at right angles to the repair

(a) (b)

Refacing of a member in bending

This would normally occur when a post has decayed and is receiving load from a brace. Two
circumstances are possible: loading in the plane of the repair joint and loading at right angles to
it (Figure 8.8).

Loading in the plane of the repair is similar to the condition described in the previous section.
It might be possible for either the original timber or the repair timber to carry the bending
moments alone but it is clearly preferable that they act together. This can be ensured by making
sure that load at the mortice is distributed into both members by providing adequate fastenings
between the two to ensure that they both deflect together.

Loading at right angles to the plane of the repair might occur when refacing a main post that is
loaded by the brace of the cross frame (Figure 8.8(b)). The bending moments might be small
enough that the sum of the bending resistances of the two timbers acting separately will be
sufficient. If the loading is in the same plane as the repair joint (Figure 8.8(a)), one will need to
consider whether it is the repair timber or the original timber, or possibly both, that is receiving
the load from the brace.

New beam ends

This is such a common circumstance that it needs discussing in some detail. Typical situations
are the replacement of the ends of floor joists and beams, roof rafters and tie beams and repairs
to wall plates. Common ways of doing this are

(a) to have a scarf or bridle joint between two timbers

(b) to fasten steel plates to the side of the original timber where the decayed material has
been replaced
(c) to have a steel flitch between the two timbers.

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

If mechanical fasteners are used, a similar calculation is used for all these methods to determine
the forces on the joint. If the member is simply in bending, then the forces at either end of the
joint will be the same, being simply the moment in the member divided by the length of the
joint. This situation can occur in a roof tie beam where the principal rafter is not directly above
the wall plate, thus producing bending in the beam. This assumes that there is no other load on
the tie beam except its own weight, which might possibly be ignored. Then, between the
principal rafters, the tie will have the tension force combined with a bending moment and there
will be no shear force to be considered. However, most members, such as rafters and floor
joists, carry a uniformly distributed load, while wall plates that have a series of rafter loads on
them can be treated similarly, that is, as if the rafters produced a uniformly distributed load. In
these members, a repair joint has to transmit both a shear force and a bending moment, which
means that the forces at each end of the joint will not be the same.

There are two possible arrangements.

(a) The uniformly distributed load on the member is carried by the original timber across
the length of the joint.
(b) The uniformly distributed load is carried by the repair material across the length of the joint.

Figure 8.9 shows free-body diagrams for both these conditions. The force at the end of the
repair, Ra, is known because this is simply the support reaction for the member, wl/2. The load
along the member is known, and there are two unknown forces at either end of the joint. These
unknown forces can therefore be found by first taking moments about one of them. In setting
up the equations in Box 8.1, it has been assumed that the actual length of the repair is the same
as the effective length, that is, the distance between the fasteners, or groups of fasteners,
although this introduces a slight error.

Figure 8.9 Beam end repair forces depending on the parts carrying the superimposed load

a b c

a b c

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

Box 8.1

Referring to Figure 8.9

Assuming a load intensity of w/unit length and a span of l, then the load on the member = wl
and the reaction at each end = wl/2.

Let the effective additional length to the end of the beam be ‘a’, i.e. the distance between the
end fastener or fastener group and the support point of the member.

Let the effective length over which the repair is made be ‘b’, i.e. the length between the fas-
teners. This will be a little less than the actual length of the repair because of the addition of end
or edge distances.

If the loads on the fasteners are P and Q, we can take moments for the length a + b about Q.
(i) Raða + bÞ = Pb + wa b + a then Q = wa + P − R
w ða + bÞ2
(ii) Raða + bÞ = Pb + then Q = w ða + bÞ + P − R

Figure 8.10 Beam end repair forces depending on the parts carrying the superimposed load

While this has been shown for steel plates fastened by bolts or screws, as in Figure 8.10, the
equations will also apply to other repairs, such as those to a scarf joint.

Scarf and bridle joints

If a full depth section of new timber has to be used to replace decayed material then a basic
bridle joint might be used (Figure 8.11). The advantage of this joint is that it involves the least
loss of original timber and is a useful joint if the timber is wide enough to allow its use. If the
decayed end of the member is cut off, the only other loss is the material required for the slot
into which the tenon on the replacement timber is fixed. The tenon will then normally be fixed

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

Figure 8.11 Bridle joint

by bolts though the timbers, although if the joint is not carrying load, as might be the case in a
repaired wall plate, peg fixings might be used. Otherwise, the load-carrying capacity will be
limited either by the bending capacity of the tenon or by that of the bolts. The forces on the
fasteners will be as before. If a single fastener is used at each end, the design of these joints will
be largely governed by the end and edge distances of the fasteners, especially in small
members, such as roof rafters. What needs to be considered in the design of the joint is the
loaded edge distance on the bolt.

Alternatively, a very simple half scarf joint, shown in Figure 8.12 with split-ring connectors, is
structurally the simplest method of joining two pieces of timber to transmit shear and bending.
This was the form of scarf joint adopted by the nineteenth century as a means of forming long
tie beams from more than one timber, with bolts in shear transmitting the tension force. When
this form of joint is used to transmit moments, it might be convenient to use split-ring con-
nectors, as suggested in the figure. As the cross-section is halved, the moment that can be
transmitted across the joint will only be half of the capacity of the full size beam.

Figure 8.12 Halved scarf joint

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

Figure 8.13 V joint

A joint devised by the Weald and Downland Museum might be regarded as a modern version
of the bridle joint. AV-shaped cut is made in the beam and a corresponding timber let into this
and fastened with epoxy resin (Figure 8.13). As with their mortice repair, the purpose is to
retain as much of the original timber surface as possible. This also has the advantage over the
bridle joint that no mechanical fasteners are used, so that it is more ‘discreet’.

Joints like the traditional scarf joints shown in Figure 8.12 are popular in repairs because the
result is visually pleasing, so that the joint can be used where the repair is to be seen. Therefore,
it is one of the more common forms of repair in timber-frame buildings, although today used in
association with timber screws. In making the repair, the cut in the original timber is made first
and the new timber fitted to this. Tightness can be ensured in the joint by setting up the joint,
clamping the timbers together and running a saw kerf along the squint at both ends so that
sliding the timber up makes both squints tight. If this method is used, for the joint to fit
properly, the timber of the repair has to be a little deeper than the original and must then be
brought to a fit by planing once the joint has been completed. If the screws or bolts are simply
set orthogonal to the member, they will produce a component of force along the plane of the
table as they are tightened up, tending to drive the two timbers apart. They should be per-
pendicular to the plane of the table.

It is worth considering the design of these scarf joints under bending loads because of their
popularity and because their behaviour seems not well understood. Some years ago, TRADA
carried out tests on such joints,i which were assembled using a number of different fasteners in
order to ascertain their relative strength and their strength compared with the parent timber.
These tests suggest that their limiting moment capacity is equal to 1/3 of the moment capacity
of the parent timber, and so it would be prudent to limit the moment on any repair joint to this
value. However, this does not help to understand the forces on a joint for the purposes of

The results of these tests do not seem to have been published.

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

With the joint in bending, the tendency is for it to open up at one end, and this can be resisted
by screws acting in tension. These will be in withdrawal in one timber, with a compressive
stress under the washer in the other. At the compression end, it seems reasonable to use the
basic stress of the timber species to determine the length of material required to carry the
compressive force and so determine the lever arm between the area in compression and the
screws in tension. A little trial and error will be needed in the design. There are no recom-
mendations for the end distance of screws used in this way but splitting of the timber is
unlikely and 1·5 diameters is probably adequate, with the point of the screw in the deeper part
of the scarf. Rather than a concern for splitting, the difficulty lies in obtaining sufficient timber
under the head of the screw, or rather the washer, especially if it is necessary to sink it below
the surface.

For simplicity, consider a joint in pure bending – as in Figure 8.14. At one end, there will be
direct compression between the timbers over a length, which might be calculated by assuming
a uniform stress over the area indicated. The distance from the centre of this compression area
to the fasteners at the other end is then the effective length of the joint.ii In practice, members
normally have a uniformly distributed load so that the scarf will be transmitting a shear force as
well as bending and the forces at either end of the joint will not be the same (as previously) but
the principal for determining effective length is the same. Design of the joint involves bal-
ancing the length of the scarf with the load the fasteners are required to carry. Carpenters
normally make the length of the scarf 2½ to 3 times the depth of the timber and that is a useful
place to start the trial and error process.

The obvious question is which way round the scarf should be made. It may be simplest to
assume that the larger force is carried in compression because that will keep the screw length to

Figure 8.14 Tabled scarf joint showing screw position and area in compression

Some engineers might prefer to take a triangular distribution of stress but this seems unduly complicated.

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

Figure 8.15 Tabled scarf joint showing screw position and area in compression

(a) (b)

a minimum. However, there are situations where visual considerations can override structural
ones, even though structural considerations suggest that the tension side of the joint should be
placed so as to have the smaller load on it. Where the member is visible, one would, given a
choice, want to show less of the repair timber, leaving more of the original in view. Compare
then, repairs to either a rafter end or a beam or joist end, both assumed to be visible from below
(Figure 8.15). The new timber is lightly shaded in both cases and the screws at the tension end
indicated. For the rafter arrangement, ‘b’ is visually the more desirable, whereas for the beam
end it is arrangement ‘a’.

Flitch plates and reinforcing bars

When flitch plates are used for repairs, they are often fastened to the timber with an epoxy resin
adhesive. This avoids the problem of drilling and lining up bolt holes in the plate and the
timber. It is not usual to carry out calculations of the bond between the epoxy resin and the
other materials because experience shows that it is the strength of the timber that governs
the design. The alternative to steel plates is reinforcing bars set in epoxy resin. Jones and
Smedley (2000) report that when using steel reinforcing bars with an epoxy, joint efficiencies
depend on the size of the original timber, varying from 63% for 150 mm × 50 mm beams to
97% for 175 mm × 50 mm beams. The purpose of the repair is to replace decayed beam ends so
that they would be used in regions of low bending moment but relatively high shear. The
design that Jones and Smedley tested comprised steel bars set into drilled holes in what would
be the timber prosthetic, while a slot was cut in the ‘original’ timber to receive the bars
(Figure 8.16), which was then filled with the epoxy grout. Test results showed that, with joints

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

Figure 8.16 Repair with epoxy resin and reinforcing bars

250 250

'Original' timber 16mm dia. bars

150 & 175

Epoxy grout


in the region of shear stress, failure still occurred in the bending section of the beam. In other
words, the joint carried all the shear that was required.

The method of design that Jones and Smedley suggest is similar to that used for reinforced
concrete beams in which the timber carries compressive stresses only. If rods of fibre-
reinforced plastics are used, it is normal to abrade their surface to improve the bond with the
adhesive. Similarly, steel plates are also abraded for the same reason.

Strengthening members in bending

Both flitch plates and reinforcing rods are also used to improve the bending strength (or
stiffness) of beams. The most obvious need for this is when rehabilitation of a building
involves floor loads that are higher than the existing structure is able to carry safely. In these
circumstances, some strengthening of primary beams may be necessary. However, similar
repairs might be required if there is a defect at, or close to, the midspan of a beam. There have
even been cases where the depth of a beam has been reduced by the removal of timber in an
attempt to produce a level soffit after the beam has deflected. There is little choice in these cases
except to reinforce the beam in some way. If the beam is completely concealed, a wide range of
approaches might be possible, including the simple addition of steel plates or trussing on the
side of the beam.

Both flitch plates and steel plates bolted to the sides of a timber to increase its load-carrying
capacity involve the same calculations for determining the sizes required. The strength of the
steel flitch cannot simply be added to that of the beam because of the differences in their elastic

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

properties. Steel has a very much larger modulus of elasticity than timber so, for any given
deflection of the flitch beam, the stress in the steel will be very much greater than the stress in
the timber. How much higher can be seen simply by comparing their moduli of elasticity. The
modulus of elasticity of steel is 205 × 103 N/mm2. whereas the minimum value for THA oak
(the value to be used for a beam) is about 10 × 103 N/mm2. The stress in the steel will be a little
more than 20 times the stress in the oak for the same deflection. More significantly, we can
compare the relative allowable working stresses. That for steel is 180 N/mm2 and dividing by
20·5 gives 8.8 N/mm2, which is below the maximum allowable stress for this timber. Thus,
when the steel has reached its maximum stress, the stress in the timber will be below the
maximum allowed and so will no longer be carrying the load that it did before the flitch was

As a first approximation to design, it might be assumed that the steel carries all the load. If this
is not possible within the depth available and some load has to be taken up by the timber, then
one should simply apply the bending formula


As the deflections are the same, Ws/EsIs = Wt/EtIt, where ‘s’ and ‘t’ refer to the properties of the
steel and timber sections. The original second moment of area of the timber will be reduced
very little by cutting the slot and this will be a relatively simple trial and error process.

In practice, a steel plate as thin as 5 mm has been found to be adequate but as this would be
impractical 9 mm plates have been used. The result is that the plate carries more than half of the
required moment and the stresses in both materials are well below the maximum allowed. Of
course, the slot cut for the steel (and hence the steel itself) will usually be slightly shallower
than the original beam and this will be a limitation on the design.

It is not always possible to bolt the steel to the timber because of problems of access. Or, if the
timber is exposed, this might be regarded as unsightly. These are sometimes the reasons why
epoxy resin is used to bond the steel plate to the timber. If access to the top of the beam is
possible, then a simple strengthening can be made by cutting a slot in the beam and letting in a
steel T section, which is screwed down to the beam. This has the advantage of being a simple
and invisible repair. In practice, it might not be possible to obtain a T section of the ideal size, in
which case a pair of angles back to back can be used instead.

Where access to the top of the beam is impossible, the use of carbon-fibre reinforcing strips let
into the sides of the beam have be used. However, this last approach, which requires the use of
epoxy resin adhesives, must be regarded as a specialist job and will not be dealt with here.

Provision of supplementary structure

The kinds of situation requiring a supplementary structure are too varied to divide into simple
categories. They might involve complete structures or very simple additions. An example of
the former is a building in Marktheidenfeld, a small town in Franconia, Germany, formed in
the eighteenth century from two adjacent medieval buildings. In doing this, a large upper floor

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

room was created that was partly in each of the two original buildings. The result was a loss of
integrity of the roof supports. To preserve the eighteenth-century room intact and provide a
satisfactory roof structure, a complete steel frame had to be inserted. This was done while
preserving the roof timbers as found and still using the original rafters to support the roof

In that case, the original structure was relieved of its role and simply became an historic relic,
so that one might regard the steelwork as a replacement structure; the term ‘supplementary’
only being used because the original was retained. More likely are those cases where the
original structure continues to carry load, although the insertion of the supplementary material
alters the overall structural scheme in such a way that the original timber is subject to lower
stresses. In this approach, there is no attempt to maintain the behaviour of the original structure
because to do so would involve an unacceptable loss of historic fabric or historic character.
Therefore, the original structure may either remain as simply a secondary structure transferring
loads to the new primary structure that is to be inserted or will be part of a new structural
system. The new structure will, of course, have to be built with minimal disruption of the
historic fabric and this must be a primary concern in its design, the engineer discussing this
with the contractor and other members of the design team.

Fairly common examples here are the insertion of bracing to prevent racking in tall roofs, the
insertion of wind bracing to cope with transverse wind loads that are inadequately catered for
by the original structure and the strengthening of roofs with inadequate purlins or wall plates.
In all these examples, the need for remedial work will be apparent through deformations in the
structure, which, while not yet fatal, might result in severe damage or even collapse if left
unchecked. This may eventually be caused because of secondary forces arising as a result of
the deformations, because of the eventual weakening of the structure through decay or damage
in an exceptional storm – or possibly a combination of these. Of course, there are many cases in
which supplementary structures have already been inserted.

A fairly commonly occurring situation involves inadequate purlins or wall plates in barns. In
some eighteenth-century barns, where softwood was used in place of oak, this was done with
no corresponding increase in member sizes. One sometimes finds structures with excessively
long bays, possibly in an attempt to economise on timber, or structures where the original
covering has been replaced with something heavier. Wall plates as well as purlins may be too
small where the weight of the roof is carried by the wall studs but where the magnitude of the
outward thrust has not been recognised. When originally built, the stresses in these members
may have exceeded those that would be allowed by modern codes with the only signs of
distress possibly being large deflections. These may have increased over the years as a result of
creep, and with time and decay the structure may now be in a critical condition. Of course, this
process is accelerated if the deflections are sufficient to compromise the ability of the covering
to exclude rain.

In such cases, one method of repair is to provide some form of supplementary structure, while
at the same time possibly bringing the structure back towards its original geometry so that the
roof can be made to work properly. If the latter is required, at least partial removal of the
covering will be needed. The wall plate and purlins could be stiffened by adding additional

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Decay and repair

timber or by reinforcement with steel plates. However, it is also possible to strut and tie the
purlin and plates at midspan, so reducing the bending moments.

Three examples of the use of supplementary structures are provided in Chapter 9. In one case,
the original structures that were contemplated were abandoned in favour of a much simpler
arrangement. In the second, the original scheme proposed was deemed unacceptable by the
architect and a more sympathetic arrangement sought. The third example involved altering the
original support arrangements.

Inadequate ‘original’ structure

Often the original design is found to be wanting in some way. It is not just that purlins might be
undersized and show severe creep deflections, a common feature of medieval domestic con-
struction, or that lap dovetails shrink and allow outward movement of wall plates and splitting
of the post jowls. In neither of these cases, or in some others where there has been poor design,
is it necessary to undertake any remedial work, because the structures have survived and will
continue to do so. The cautionary tale of Chapter 5 is a case in point. Difficulties arise,
however, when the engineer is called on to carry out remedial work and is unable to justify the
original structural scheme, that is, to demonstrate that the structure will be safe unless some
new structural scheme is introduced.

Previous repairs
In many cases, the conservator will be confronted with a structure that has already been repaired
in the past and such repairs may be considered to have significance as part of the history of the
building. They are, of course, indications of the inadequacy of the original design and so provide
some insight into the structural knowledge of both the original builders and of those who made
the repairs. Some will be adequate repairs that can be left to do their job, some may have proved
inadequate or have become inadequate as a result of further changes in the structure and so will
require new structure, some may be little more than inadequate bodges. For all of these situa-
tions, difficulties may occur if the presence of the earlier repair compromises the design presently
being considered. It seems sensible to make a judgement about the value of a historic repair in the
same way that one would make a judgement about the value of any other aspect of the historic
fabric. This, of course, means bringing to bear some knowledge of the history of structural design
as it applies to this kind of structure so that the historic significance of the work can be assessed.
If the repair is simply an inadequate bodge, and many are, there seems little point in retaining it if
the present work would be better without it.

Baird JA and Ozelton EC (1984) Timber Designers’ Manual, 2nd edn. Granada, London, UK.
BSI (1984) BS5268-2:1984. Structural use of timber. Code of practice for permissible stress design,
materials and workmanship. BSI, London, UK.
BSI (2002) BS5268-2:2002. Structural use of timber. Code of practice for permissible stress design,
materials and workmanship. BSI, London, UK.
Forest Products Laboratory (1987) Wood Handbook: Wood as an Engineering Material. Forest
Products Laboratory, Madison, WI, USA.
Harte AM and Dietsch P (2015) Reinforcement of Timber Structures, A State-of-the-Art Report.
Shaker Verlag, Aachen, Germany.

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.
Repair of Historic Timber Structures

IStructE (Institution of Structural Engineers) (2007) Manual for the Design of Timber Building
Structures to Eurocode 5. IStructE, London, UK, p. 109.
Jones R and Smedley D (2000) Replacement of decayed beam ends with epoxy-bonded timber
composites: structural testing. The Structural Engineer 78(15): 19–22.
Lavers GM (1967) The Strength Properties of Timber. Forest Products Research Bulletin 50.
Building Research Station, Watford, UK. (Reprinted 2002.)

Downloaded by [ La Trobe University] on [21/06/20]. Copyright © ICE Publishing, all rights reserved.