Appraisal and repair of St Peter’s at the Cross, Chester

A_CI

area of cast iron, mm²

e

eccentricity of the thrust-line in the masonry, mm

f_b

compressive strength of masonry unit, N/mm²

f_j

compressive strength of mortar joint, N/mm²

f_m

masonry compressive strength, N/mm²

H_R

horizontal thrust reaction at springing hinge, kN

M_c

moment about the crown hinge, kNm

R

linear thermal strain, mm

t

thermal range, °C

V_R

vertical thrust reaction at springing hinge, kN

W_n

weight (including superimposed load) of respective slice of masonry, denoted by integer ‘n’ kN

x_n

distance of respective slice of masonry centroid from crown hinge, m

α

coefficient of linear thermal expansion, ×10⁻⁶/°C

δ

deflection, mm

σ

compressive stress in the masonry, N/mm²

The church of St Peter’s at the Cross dates from 907 AD, with the present building thought to date from the fourteenth century, with major restoration phases recorded in the seventeenth, eighteenth and nineteenth centuries. It is a Grade I listed building (Historic England, 1955). The structure principally comprises a traditional stone masonry skeleton. It has two naves, two aisles, and a tower to the west end. The nave upper walls and the respective bays of lead-clad timber roof structure bear on a series of arcades, each of which are three bays long. The arches are of varying pointed forms; the spans to the west are the shortest with deepest rise, and the spans lengthen to the east where they are of proportionately shallowest rise for the span.

The arcades are not externally buttressed at either gable (Figure 1), and a major outward lean is exhibited by the east gable. The north arcade had been cross-tied with cast iron tie rods and pattress plates gable-to-gable, as part of the nineteenth-century repairs.

The central and southern arcades did not receive such remedial work in the nineteenth century. They are buttressed to the west end by the heavy tower and abutting terrace, but the east gable is plain and unbuttressed. During a due-diligence structural condition appraisal as part of a planned reordering project, a major outward lean of the east gable about the central arcade was observed.

This appraisal data told of badly resolved thrust between the arcades and the self-weight of the gable, and a very delicate state of equilibrium: a serious structural condition was determined. Intrinsic deficiencies of this magnitude between shape and weight in structures of this age are rare in the authors’ experience. This article records the appraisal, structural assessment, design and implementation of temporary stabilisation and permanent repair works, including the load transition between the pre-loaded temporary works into the permanent works.

The starting point of structural appraisal is to first understand structural form and composition (Beckmann and Bowles, 2004; Institution of Structural Engineers, 2010). The church is broadly square on plan (some 21 m × 20 m), with a stocky tower to the west end, where it abuts a terrace of buildings. The perimeter walls are around 12 m tall, some 900 mm thick, and comprise traditional stone masonry walling built of two ashlar facing skins of Chester red sandstone bedded in hot-mixed lime mortar together containing a rough rubble core. Internal piers are through-bonded ashlar work. Aisles measure 3.8 m wide, and the two naves measure some 5.4–6.0 m wide each. Low-pitch lead-clad timber roof structures span the naves and aisles. The nave roof structures principally comprise cusped or trussed-up beams spanning between wallheads. The roof affords good lateral and plan restraint to the nave flank wallheads, although no connectivity with the gables was identified.

The nave wallheads carrying the roof bear on a series of three arcades, each of three spans, of varying span-to-rise ratios (Figure 2).

The north arcade is not buttressed at either gable, but is cross-clamped by two cast iron tie rods brought through the centreline of the wall to large stiffened cast iron pattress plates at the outer faces of the west and east gable walls (Figure 3). These works seem to date from the nineteenth century based on form and materiality.

The central and southern arcades did not receive such intervention during that repair campaign. To the west end, the stocky, heavy tower provides a buttress to each arcade, and this part of the west gable abuts adjacent structure (although this was not always the case over the life of the building, with the church recorded on historic maps as a detached structure, see British History Online (2005). To the east end, there is a plain masonry gable, with stained glass arched window openings channelling the wall into piers that align with the arcade positions. Historical review indicates the presence of a pentice structure wrapped around that gable, which was demolished in 1803 to admit the widening of the road. The possible presence of former external buttresses hidden by the pentice is not known (Figure 4).

Foundation depth, detail, and geotechnical properties of the formation medium are not known. The church is built on the site of a former Roman praetorium (Hartwell, 2011), and central Chester has much mediaeval fill. British geological survey boreholes 100 m northeast and 500 m south of the building record deep made ground to between 2.8 and 5.0 m below ground level above the sandstone bedrock. A church has existed on the site since the tenth century, and re-using of earlier foundations is a possibility. No conclusive information on the current building foundation yet exists; however, written records of an undercroft to the north aisle (since infilled, and not re-excavated) may indicate deep substructure. This puts foundations bearing near or on weathered bedrock as a possibility, but not proven.

The principal findings of the structural condition appraisal are outlined below:

• The east gable suffered an outward lean of some 155 mm over its lower half, principally affecting the piers formed by openings in the east gable at the end of each arcade.

• The outward lean of the gable in section about the arcades drives an associated bow on plan between the flanks by the same amount.

• Apparent sag in the crowns of each of the east bay arches, with downward tilting of the bed-joint coursing in the spandrels over the affected arches, tending down towards the dropping crowns.

• Sharply defined separation cracks up the spandrel/gable wall joint at the central arcade, stemming from nil thickness at springing coarsening with height to some 40 mm thick at wallhead.

• A thick gap of some 50 mm between the edge of the roof structure and the inner face of the east gable.

• Skin separation of the east gable about the central arcade springing was detected up the jambs.

• Fine cracking through the nineteenth-century separation crack repair (notionally toothed indent work) to the spandrel/gable joint at the tied north arcade was also detected.

• Slight separation between the arch extrados and the ashlar spandrel to the east span of the central arcade was detected.

• A heavily eroded band of masonry at the base of the outer face of the east gable was observed, eating into the profile of the wall.

Figure 5 presents an exaggerated record of the articulation detected in the structure. At the time of the initial appraisal, no measured survey had yet recorded the shape of the building.

The above appraisal data supported a diagnosis of a very delicate state of equilibrium between the thrust of the nave arcades and the shape and weight of the east gable. A serious structural condition was determined.

A possible timeline of the above structural deterioration of some 150 years was considered. The demolition of the pentice in 1803, and the possible removal of buttresses hidden by the pentice itself, seems a likely trigger for the articulation of the arcades. Judging from the lack of intervention in the nineteenth century to the central and southern arcades, this suggested that there was not significant deformation or distress to these arcades at the time of those works; a linear development of structural movement of some 150 mm spread over 150 years was supposed, giving a crude estimation of some 1 mm/year spread; the situation was serious, but did not feel acutely dangerous.

A measured survey of the building by point cloud to record its deformed shape and inform a calculation-based assessment of thrust equilibrium was instructed.

The east span of the central arcade was considered for assessment, initially as a half-arch, carrying the masonry overhead, and the appropriate bay width of roof. This dead load was then combined with a live load of snow on the roof, or a severe suction force on the gable during a gale. Historic church accounts tell of money paid for ‘throwing snow off the roof’ (Simpson, 1909). The combination of heavy drifted snow on the roof and severe suction force on the gable at the same time was considered unrealistically onerous during the period of assessment pending repair. A thrust-line equilibrium assessment using the ‘safe masonry theorem’ was undertaken (thrust-line safe masonry theorem uses a funicular polygon based on graphic statics to arrange force vectors into a scaleable diagram, then used to plot the forces down through the structure; see Wolfe, 1921; Hendry, 1998; Heyman, 1982, 1997). Figure 6 presents the working as follows:

The thrust-line assessment was considered about the base of the wall at the lower external ground level, as no detailed information about underground masonry, foundations, or ground conditions was available or achieved in the project. A shallow spread foundation formed on mediaeval fill would be expected to have rotationally failed and collapsed the wall with such an eccentric thrust position (giving a grossly over-stressed peak of pressure to the outer edge of such a foundation, on a very weak formation medium). Accordingly, a tentative assumption of deeper substructure masonry bearing on comparatively shallow bedrock was made during the assessment. This would allow the formation medium (rock) to tolerate a much higher contact pressure than the fill, and potentially explain why the building has stood for so long without such incident. With the unknowns remaining, no further speculative analysis is given in the calculations.

The calculation-based assessment gave a thrust at the springing of some 120 kN horizontal force component, and 250 kN vertical force component. The stability of the gable was assessed in a similar way (Huerta, 2010). The effect of the gable arches channelling the weight of the gable peaks into the piers formed by openings improved the available weight in the wall to resist this thrust (Figure 6(b)). About origin ‘O’, considering destabilising moment against restoring moment allowed the position of thrust at the base of the gable to be located, determining that it was very close to the outer face of the wall; under working loads, the thrust was some 35 mm in from the extreme outer face (Equation 1). In the event of a severe suction force, the thrust-line essentially struck the edge of the wall, on the point of equilibrium. The effect of this gross eccentricity was being exacerbated by aggressive stone decay from soluble salt damage to the outer face, slowly eating away at the residual geometric factor of safety.

Equation 1: Assessment of gable stability

This assessment confirmed a very delicate state of equilibrium under realistic loading conditions. To better understand the working margin of safety against instability and risk of disproportionate collapse, potential secondary load paths were looked for which would need to be exhausted before collapse were to actually occur. These would need to be mobilised by further significant deformation, giving warning of further distress. The presence of the stiff ashlar spandrel walls over the voussoir arches was introduced into the assessment, to see whether the horizontal force component of thrust into the gable could be reduced.

If arching were to occur in the stiff ashlar spandrels over the arcade arches, significantly more leverage could be mobilised to markedly lessen the horizontal thrust component at the springing (taking some 120 kN down to 90 kN, further reducing to 80 kN under dead-load only conditions). This may reduce horizontal thrust by about a third, which would bring the corresponding thrust-line in the gable to around 200 mm in from the edge of the wall, and build in some resistance against wind suction loads (accounting for spandrel arching and sheltering effects of adjacent buildings may explain why the gable had not blown flat historically).

The ring separation observed between the arch extrados and the spandrel panels at the east span would mean no longitudinal shear flow from arch proper into spandrel (as is relied upon by masonry bridges with concrete or masonry backing to create a composite deeper arch – this approach allows the thrust-line to escape the masonry arch into the backing, provided it is a strong and stiff enough load path, relying on longitudinal shear flow to take the thrust out of one element into another, in similar manner to multi-ring arches, see Harvey, 2012). However, a possible structural behaviour of the arch carrying its own weight, generating its own dead-load thrust at the springing, and then the stiff spandrel arching separately with its increased leverage and reduced horizontal thrust, lifting off the more deformable arch below was imagined (Figure 6(a)).

The diagnosis of ‘delicate state of equilibrium’ would still apply even with potentially redundant load paths conceivable within the masonry, as robustness against disproportionate collapse was grossly deficient: the appraisal and assessment together indicated a structure flirting with equilibrium if fully loaded by wind. The gable abuts a busy road; the consequences of failure would be catastrophic (the gable would fall onto the road and the rows opposite, while the arcades would sway to the east and the roof would collapse into the naves).

The thick, sharply defined separation cracking indicated continued movement of a progressive nature (the further the wall leaned, the lesser its restoring moment potential, and an increasingly sagging crown steadily increases the horizontal force component of thrust).

As a point of note, despite this serious diagnosis, there was no correlation between compressive stress in the masonry and the overall structural pathology. The peak compressive stress in the loaded gable near the base of the wall remained within the allowable compressive strength of the masonry composite: the wall was not grossly overloaded. Figure 7 presents a strength curve of traditional masonry built with lime mortar, using the corrected formula developed by Costigan et al. (2015), which was used to check the working stresses. This ‘high load, low stress’ outcome is characteristic of historic masonry structures, and informs that it is force position, not force amount, that has the direct bearing on structural pathology.

The prospect of major permanent repair was some way off (the church is an independent charity, with very limited capital funds; with the Grade I designation this would mean that both Faculty and Listed Building Consent would be required to admit its permanent repair). A temporary holding repair was therefore designed to arrest the movement, and buy time to better understand the building and allow fund raising to cover the capital works.

The goal of temporary works was to provide a full-strength ‘seatbelt’ which would be strong enough to tie the full thrust of the nave arcade if it had to (if mobilised by significant further outward spreading of the arcades/leaning of the gable wall), and be stiff enough to attract the load up front without undue further deformation of the structure.

In a tension system, this would require the temporary works to be pulled taut, requiring a pre-load. This pre-load would be beneficial for the internal resolution of forces within the gable, pulling the thrust-line inwards away from the outer face of the wall (improving the geometric factor of safety) and easing the eccentric pressure profile in an as-yet-unknown foundation.

The temporary stabilisation work would need to be reversible (ICOMOS, 2003), and not damage the building at the bearing surfaces. Critically, it would need to allow for the insertion of the permanent works without clashing and allow for the pre-load to be worked out of the temporary works into the permanent work assemblies, while maintaining the stability of the building.

Lifting chains were selected for strength and stiffness, pulled taut by a turnbuckle tensioner. Around piers and anchorage points, the chains were replaced by lifting slings, choked off once aligned (Figure 8). Oak baulks formed temporary pattress plates to distribute the contact pressure at the pinch-points. Load cells were built into each system, which gave proof of pre-load and would give insight as to further structural activity (increasing force in the system would imply further arcade spread – a ‘tightening of the seatbelt’).

The level of pre-load was determined to be 20 kN for the central arcade, intended to draw the thrust-line in by some 100 mm from the outer face of the gable (resolving the moment expression in Equation 1 with the increased restoring moment of the 20 kN tie at 6.3 m height above origin ‘o’, raising M_Rst to some 1552 kNm), and achieve enough system stiffness to allow for immediate tie mobilisation in the event of a severe suction force on the gable, or snow on the roof. The anchorage points were the springing at the gable (the stained glass was removed by conservators), and the base of the tower pier. For the south arcade, which exhibited less onerous structural pathology, the lesser value of 10 kN was adopted, as the presence of the gallery precluded an anchorage point at ground level, and was thought to provide some limited tying capability itself. Figures 9 and 11 record the temporary stabilisation works:

Limited strain data was available for the lifting slings; 7% stretch at the full working load limit was recorded in test data, and a linear stress–strain relationship was assumed but not confirmed. During tensioning, the straightening of the chain system and the bedding in of the chokes took up much turnbuckle tensioner thread. The 600 mm tensioner had to be uncoupled and reattached to the tightened chain several times to remove all the slack from the combined system.

Once taut, the system was locked off and a load monitoring regime began. In the first 3 months of the temporary stabilisation system, a steady drop in the force in each system was observed. Each system was tightened to the full design pre-load when the load dropped below 80% of the design load. However, on re-survey of the shape of the building by point cloud, an inward movement of the gable of some 5 mm at the central arcade was detected. This inward displacement under a load far less than the outward thrust exerted on the gable was surprising, and further application of tension in each system was suspended, and the temporary works allowed to settle naturally from that point. A resting load of some 16 kN for the central arcade and 7.5 kN for the southern arcade was then sustained over the 2-year life of the temporary works, with small changes in load then attributed to thermal movement of the building and the temporary works system itself.

A possible explanation for the observed inward pull of the leaning gable under a small load compared with the outward thrust may be the regime change between a grossly eccentric pressure profile in the masonry (either within the gable wall itself or at the foundation) to the eased stress profile for more concentric thrust. Open bed joints to the inner face of the leaning gable were never observed during the appraisal, but considering the possibility of deep foundations and an infilled crypt, it is possible that such a phenomenon did occur but was not witnessed. With the significant reduction in thrust eccentricity, a potential closing of those joints to the inner face (at or below ground level) could account for the inward movement of the wall further up (Figure 10).

It also seems likely that the 20 kN inward pre-load was pulling against a reduced outward thrust (80 kN not 120 kN) taking account of possible spandrel arching as plotted in Figure 6(a), contributing to the apparent inward movement observed.

With the building stabilised, attention turned to exploring options for permanent repair. A thorough calculation-based assessment of possible thrust resolution within the masonry skeleton was undertaken, dealing with the arcade as a whole, to better understand structural behaviour in the round (Figure 12). Broadly speaking, this confirmed poor resolution of forces: put simply, the gable needed external buttresses. In the authors’ experience, such gross intrinsic deficiency between shape and weight is rare to find in a 700-year-old building.

Each arch was sliced into weight increments, with roof or vault loading superimposed on these segments. Taking moments about the crown hinge (which was placed as close to the extrados as possible and its position iterated side-to-side to achieve the minimum thrust conditions pushing against the gable), a series of equilibrium expressions could be found. Once they were solved for H and V force components, the funicular polygon could be drawn and the thrust-lines plotted through the masonry (Equation 2).

Equation 2: Equilibrium assessment for each arch

The addition of external buttresses was ruled out as a possible design solution owing to the layout of the highway, and the visual change to the building with such an intervention in this context would be major. The obvious solution was to cross-clamp the central and southern arcades just as the Victorians had done the north arcade (Figure 13). Assessment of the nineteenth-century repair gave the following:

• Total thrust from the arcade to be resolved was around 120 kN.

• Tensile strength of two 40 mm dia. cast iron rods (A_CI = 2 × 1250 mm², with a permissible tensile strength of 23 N/mm², see Bussell, 1997) ≈60 kN.

With further deformation of the wall to mobilise the full 60 kN permissible tension in the bars, the thrust-line could be pulled essentially concentric within the gable, implying a high level of engineering finesse on the part of the builders at the time.

The nineteenth-century tying works essentially took a spreading arcade and converted it to ‘a wall with holes in it’, deleting the horizontal thrust component from the gables. Appraisal of this intervention some 150 years on validates the success of this approach, and the authors sought to replicate this in the present repair.

To mitigate the impact of the tie rods on the use of space within the nave (and avoid a clash with the altar), a single rod was selected. The tensile strength of cold worked stainless steel is significantly higher than cast iron (700 N/mm² proof against 92 N/mm² ultimate, see BSSA, 2025), and so a single 24 mm dia. rod would achieve a full-strength intervention that could triangulate the full thrust of the nave arcade if required.

A complication to cross-clamping gable-to-gable for the present repair was the presence of an abutting structure to the west, and the impact on the appreciation of the mediaeval tower crossing, its vault and the wall painting on the piers, if the tower was to be traversed with tie rods. Using thrust-lines, the capacity of the heavily loaded, thick piers at the east side of the tower was checked for use as an anchor point for the tying system. This demonstrated ample capacity to accept this tie force (Figure 14):

Pattress plates (see Bussell et al., 2003) were sized for stiffness and to replicate the character of those employed by the nineteenth-century repair (Figure 15). The plates were designed by assessing each half cantilevering out from the central tie, and bending deflection under full load was limited to L/350 at the tip of the ‘cantilever’ under full-design loading (achieving L/1000 stiffness under the 20 kN pre-load). The internal plates were designed to be much smaller than the external plates (even though their function and loading would be the same, to improve visual scaling; the outer plates are viewed from further distance). The stocky through-bonded ashlar pier meant that bearing strength and pull-through resistance were comfortably accommodated by the structure.

The tying works were intended to be tensioned at the middle of the UK thermal range (+10°C) (Building Research Establishment, 1979), to minimise the effect on the masonry skeleton of thermal strain in the metalwork in the long term. By tensioning the bars in the middle of this range, the potential thermal strain would be half that of either extreme. Thermal strain was predicted using Equation 3 as follows:

Equation 3: Estimation of thermal strain in the tie rod.

In practice, the 17-m long tie rods are internal, and the church is heated to prevent the temperature dropping below 10°C all year round, and with its thick walls the internal temperature barely rises above 16°C, giving a more realistic thermal range of around 5°C–10°C, reducing the potential thermal strain to the order of a few millimetres in practice.

The permanent works were cored concentric through the arcades by way of rotary coring rigs, bolted to stiff scaffolding. Alignment with the pattress plates was confirmed by laser. The ties were fitted straight and hand tightened (the flexible rods needed to be suspended on scaffolding to get them straight without pre-loading). The tension in the temporary works system would then need to be worked out into the permanent works, while maintaining an overall pre-load on the gable (to maintain the improvement in geometric factor of safety), but without increasing the total force in the system. This is similar to a ‘rope-to-rope transfer’ operation in rope access work. A load transition methodology was designed, with a ceiling of 20 kN for the central arcade and 10 kN for the south arcade (Figure 16):

The load transition was managed by a hydraulic bolt stretcher (‘Hydrajaws’) on the permanent works assembly, and by way of the turnbuckle tensioner in the temporary works system (Figure 17). Each gave real-time proof load measurement, which was monitored throughout the transition. Over the course of a day, both systems were transitioned.

After the load transfer, the spandrel/gable wall separation cracks were then pinned with flexible deep pinning ties (including the re-cracked nineteenth-century repair at the north arcade, which was poorly bonded), and the interface grouted with traditional hot lime grout to achieve deep crack filling but without the brittleness associated with hydraulic/cementitious grouts (Copsey, 2019). Flexible deep pinning ties were opted for over thicker sock anchors, to avoid over-stiffening the bond and potentially changing arching behaviour in that area of the structure, mindful that the thermal strain potential of the tie rods will continue to require some deformability in the structure overhead.

Force position, not amount, had the direct bearing on structural pathology, and this is affected by the centroid of the masonry element in question (the centre of gravity of a leaning wall affects its restoring moment against destabilising forces). Therefore, for other churches and similar traditional masonry buildings, a primary concern for the surveying engineer is detection of change in shape. Quinquennial surveys of church buildings should attempt to understand changes in building shape, and combine this with detected symptoms of structural distress, to give an earlier indication of structural activity.

The authors use the same appraisal and assessment approach outlined in this article to interpret the structural significance of articulated masonry in other buildings. In contrast to the present work at St Peter’s, the arcades at Lancaster Priory suffer similar outward spreading to the east, where the chancel gable leans outwards by some 165 mm over its 16 m height, with alarming deformation to the slender piers. Using thrust-lines, stable load paths have been demonstrated in spite of the deformed shape, and allowed a better interpretation of the massive buttresses to the eastern gable as almost certainly remedial in response to this spreading (Figure 18). This allowed the authors to justify no new structural intervention was necessary – a vitally important part of structural conservation engineering practice.

Figures 19 and 20 present the completed repair of St Peter’s, which was completed on time and in budget, with generous support from the UK Shared Prosperity Fund and the National Churches Trust, and other supporters.

• A fourteenth-century church was found to suffer a major intrinsic deficiency between shape and weight – a very rare find in such an old building. A plausible explanation is that the demolition of the pentice and widening of the road removed external buttresses or otherwise buttressing structure from the east gable.

• Nineteenth-century cross-clamping works resolved this intrinsic thrust equilibrium deficiency at the north arcade, and this provided a good blueprint to replicate in the present arcade stabilisation works to the central and southern arcades.

• A thrust-line equilibrium assessment was used to estimate the thrust amount and position in the leaning gable and evaluate the effect of the proposed triangulation work.

• Inward deformation during tensioning of the temporary works system occurred at a far lower load than predicted by assessment, and the authors attribute this to the possible closing/squashing of the mortar joints on the inner face of the wall as the thrust eccentricity reduced, moving into the middle third of the wall.

• Force position, not amount, had the direct bearing on structural pathology: a structure flirting with equilibrium, in a serious condition, still had very low internal compressive stresses when compared with the strength of the masonry.

The authors would like to thank the professionalism of the contractor, Recclesia, the generous funding support provided by donation, and the Vicar Jonathan Phillips for all the encouragement and leading his building group team with such enthusiasm.