Thursday, April 28, 2005
the butcher report
REPORT PREPARED BY G.W. BUTCHER
TE RATA SUSPENSION BRIDGE
Item 10 TOR
"Was the design adequate for the bridge's intended use?"
Since the collapse of the bridge involved only the timber deck structure, discussion of the design is restricted to those elements.
Copies of the construction drawings for the bridge cannot now be found. Details of the bridge deck structure however can be determined from, the design calculations (undated) prepared by Lt JR Armstrong RNZE, the RNZE Quantity Survey Sheets, the photographs in the Works Consultancy Services Investigation and Report of 5 Apr 94, the photographs held by Mrs Berryman, and by inspection of the small amount of salvaged material still available on site.
I understand from Mr Berryman that the final design prepared by Lt Armstrong was subject to review by Mr Berryman and Mr W Emmett a Wanganui Bridge Contractor and some changes were made to the towers, cables, and the connection of the cables to the anchorages. The decision was also made at about this time to use second hand Oregon from the Imlay freezing Works and available from the Wanganui Demolition Co. for the construction of the timber deck structure. This was in place of the more expensive treated Pinus radiata allowed for in the earlier designs prepared by SME. Totora was however used for the decking.
There is nothing in the design calculations document supplied that would suggest that the calculations have been subjected to a checking protocol, or QA procedures, or a peer review.
The final design of the bridge was apparently carried out in the period 1985 to 1986 and the bridge completed on 22 Mar 86 when the first vehicle used the bridge.
The design of the structure, apart from the wire ropes, was based upon the use of allowable or working stresses for the main structural elements. Wheel loads were regarded as point loads for the design of members which is conservative but in accordance with design procedures given in the references.
References covering design procedures and material properties are listed on the first page of the document and these include RE and Australian Army manuals, an SME Bridging Wing Precis on suspension bridges, and an FRI publication on the strength properties of New Zealand-grown timber. There was no reference to applicable New Zealand standards such a NZS 3603:1981 "Code of practice for timber design" nor to documents such as the "Bridge Manual" 1956 and the "Highway Bridge Design Brief", CDP701 1978 of the HWD. The latter documents set out the recommended practice for the design of bridges in New Zealand at that time.
There were several technical errors of judgement in the design calculations, two of which involved the design of the transoms and by good fortune were self cancelling. The errors were:
(a) The allowable stresses used in the design of the timber members were far too high for the grade and timber species used. The following table compares allowable stresses for Douglas fir beams with moisture contents greater than 25% ("green") and with the basic stresses modified where appropriate with the short duration load factor k1.
|Used in Design||NZS3603 (k1=1.35)||MLW Vol 4 Pam 1 (k1=1.25)||US Timber Con Manual (k1=1.15)|
|1. Bending (MPa)||18.75||9.18||10.75||8.86 - 10.56|
|2. Shear (MPa)||6.27||1.08||1.08||0.66|
It can be seen that the allowable stresses used in the design calculations are between 1.75 and 2 times higher in bending and 5.8 times higher in shear that permitted by MLW Vol 4 Pam 1 and the New Zealand timber design code current at the time. Similar discrepancies apply to allowable stresses in Totara.
The effect of the use of such high stresses in design is to reduce the load factor against failure to almost 1. Such a low value is unacceptable in structural design although in this case there are some compensating factors which would increase the real load factor. For example, the design allowed for an impact factor of 15% which follows the procedure set out in ME Vol III Part 1 for timber deck elements apart from the decking. Australian and New Zealand timber design practice on the other hand recognises the increased resistance of timber to transient and short duration loads and does not require an allowance for impact. See MLW Part Two Vol 4 and the Bridge Manual. The nett effect on the design was to increase the real load factor by 15% in the value of the stringers.
The reasons why the allowable stresses adopted for the design were so high are:
1. The designer used the FRI Bulletin No 41 without adjustment to determine allowable stresses in bending and shear. Even though the lowest recorded test values were adopted this was an error of judgement. The stated purpose of the bulletin was to record the strength properties of small clear specimens (20mmx20mm) of New Zealand-grown timber determined from mechanical testing carried out over a number of years. It is pointed out on page 2 that the results at best are only a comparative guide to the selection of any timber for strength. The Bridge Manual suggests that for such test results the basic working stress in bending should be one fifth of the modululus of rupture and one seventh of the shear stress at failure. If these reduction factors are applied to the mean strengths given in Table 2 of Bulletin No 41, the basic stress for Douglas fir in bending would be 8.94Mpa and in shear 0.87 MPa. These stresses are reasonably close to those given in MLW Vol 4 Pam 1, the New Zealand timber code and the US Timber Construction Manual.
2. The information contained in Bulletin No 41 has been misinterpreted and the compression strength parallel to the grain taken as the allowable bending stress. The strength value that should have been adopted and reduction factors applied, is the modulus of rupture. The compression strength is usually less than half the modulus of rupture. These aspects of the structural strength of timber are discussed in some detail in ME Vol III Part 1 Basic Bridging 1981 and MLW Vol 4 Pam 1.
(b) The timber deck acts as a grillage and the wheel loads cannot be distributed uniformly to all the stringers or bearers as has been done in the design calculations. The design references ME Vol III Part 1 and MLW Vol 4 Pam 1 set out the procedures to be adopted. These involve the use of distribution factors which are dependent on the relative stiffness of the deck as a grillage but are about 1.6 for the initial design of single lane bridges.
The effect of neglecting these requirements is to reduce the load factor against failure of the stringers.
(c) The origin of the expression for determining the load is the hangers or suspenders given on page 9 of the design calculations is not clear. The live load calculated for one hanger is greater than the axle load which is not physically possible. On the basis of the greatest eccentricity of load on a transom it can be shown that the hanger load is only 65% of the value of 29.6kN (3.014 tonne) calculated on page 10.
The hanger load was then used in the design as the reaction on the end of the transom. See page 16 of the calculations. The nett result of the hanger load being nearly double the actual value was that the high stresses used in the design of the transom were compensated for and an acceptable load factor of 1.8 against failure arrived at.
The nominal strength of the deck members may be calculated from the characteristic strength information given in the Limit State code for timber structures, NZS 3603:1993 assuming the timber was in good condition at the time the bridge was constructed. The characteristic strength is defined as an estimate of the lower 5-percentile value determined with 75% confidence from tests on a representative sample of full size test specimens. In determining the wheel or axle load capacities in the following table, the full effects of load dispersion due to the running plank and deck thickness and the tyre "footprint" have been taken into account. The load factor is the ratio of the wheel load capacity of the member to the design wheel load of 12.3 kN(1.25t).
|Deck Element||Bending||Shear||Load Factor|
|1. Decking||48kN (4.9t)||71kN (7.2t)||3.9|
|2. Stringer (As designed)||10kN (1.0t) 2.0t Axle||50kN (5.0t) 10t Axle||0.8|
|3. Stringer Composite (approx)||23kN (2.3t)||50kN (5.0t)||1.8|
|4. Transom||22kN (2.2t) 4.4t Axle||53kN (5.4t) 10.8t Axle||1.8|
At the time of the design, MLW Part Two Vol 4 would have been a good basis for the design procedure and documentation for the project. It contained relevant worked examples for elements and particularly the timber deck. If it had been followed fully on this project, a satisfactory and consistent design could have been produced compatible with the then current New Zealand codes.
In my opinion the design procedure for the timber deck structure was unsatisfactory and resulted in variable load factors and stringers that were undersized. It was fortuitous that the owner elected to install substantial running planks over the stringers during construction of the deck and kept them well nailed throughout the life of the bridge. Full or partial composite action with the stringers was thus possible and together with greater transverse load distribution provided additional stringer load capacity not taken into account in the design.
As a result, I can only conclude that the design of the deck structure could not be considered as adequate for the bridge's intended use. The inadequacies of the design procedure however did not contribute to the failure of the transom nor to the resulting collapse of the section of bridge deck.
GW Butcher 26 Sept 94
TE RATA SUSPENSION BRIDGE
Item 16 TOR
"Were the construction materials/methods adequate for the bridges intended purpose/design?"
The basic concept of the structure and the construction materials employed were in my opinion quite adequate for the design of bridge and its intended purpose. The proof of this is in the decision to repair the bridge, replace the deck structure in timber apart from the transoms which are now in structural steel, and the reuse of all the other components of the bridge. The decision to go to steel transoms probably reflects the longer life that could be expected from steel members compared to timber beams, and the difficulty of replacing such components in the future. The availability of timber of the required dimensions and additional costs would also need to be taken into account.
The decision to use untreated timber in the original bridge cannot be supported. The Oct/Nov 1985 design was based upon the use of Macrocarpa (Cupresses macrocarpa) which untreated has a reasonable durability when not in ground contact. The alternative suggested was tanalith treated Pinus radiata. In the construction of the bridge, I am informed that second hand Oregon (Pseudotsuga menziesii or Douglas fir coast region) from a building in Wanganui was utilised for the transoms and stringers. Totara, which is very durable, was used for the decking. The use of these timber species is confirmed by the Armstrong design calculations.
Imported Oregon has been widely used in New Zealand in the past as a structural material but in my experience always under cover. It is durable when continuously immersed in water, but has a very short life when exposed to the weather or when subject to alternate wetting and drying.
Douglas fir is difficult to treat using conventional water-borne preservation pressure treatment systems.
The durability of timber and its importance in bridge design is discussed in one of the references cited in the Armstrong design calculations. ME Vol III Part 1 Basic Bridging para 0327 states:
"Bridge design in timber must allow for deterioration in time, and the more so if the bridge is semi-permanent in nature. The selection of the timber available which is most resistant to the type of attack likely to prevail is clearly of importance. Provision must also be made for regular inspection for deterioration or damage."
[...possible missing text here...]
TE RATA SUSPENSION BRIDGE
Item 19 TOR
"What caused the bridge to collapse?"
The collapse of three bays of the timber deck structure on 22 March 1994, was due to the failure in bending of a timber transom when subjected to the rear axle load of 23kN (about 2350kg) of a vehicle crossing the bridge.
The transoms span between the hangers or suspenders from the main cables and support the stringers and in turn the decking. Two transoms failed in the collapse of the three bays (about 8m) of timber deck, causing the vehicle to plunge into the Retaruke River about 30m below. There was no evidence to indicate that the deck planks or stringers failed first which could have resulted in the same type of collapse experienced on 22 March. This is supported by photographs of the end of the deck remaining after the collapse, which show two sections of stringers from under the wheel tracks still intact and hanging down from where they are lapped and connected over the transom. Inspection of a transom lying on the true right bank under the new bridge, shows considerable and serious decay of surfaces which had been in contact including the internal faces of the laminated member.
The vehicle crossing the bridge at the time of the collapse was a Nissan 720 single cab 4WD flat tray light truck which was being used to transport boxes of full trays of honey from sites on the farm to the main road. Based upon the information supplied by Mrs Berryman and Nissan New Zealand Ltd, and using information in the Works Consultancy Services Report, I conclude that the vehicle payload including passenger was about 1500kg and the gross vehicle mass 3240kg. In their Fax of 11 May 1994, Works Consultancy Services suggest that with a payload of 1500kg the rear axle load was 2350kg (23.0kN). This load is very close to the design axle load of 24.5kN (2500kg) used in the Armstrong design calculations.
As designed, the transom, based on its nominal strength in bending, had a wheel load capacity of 22kN and a load factor of 1.8 against failure under the design load of 12.25kN.
The transom failed under a wheel load of 11.5kN or very nealy the design load, an almost 50% reduction in the nominal strength of the member in bending. The reduction in strength of the member was in my opinion entirely due to decay of the untreated timber.
GW Butcher ?5 Sep 94
In my opinion, Oregon timber should not have been used for the main structural members of the deck structure of the bridge when it was obviously intended to be semi-permanent or permanent in nature.
GW Butcher 17 Sep 94
TE RATA SUSPENSION BRIDGE
Item 17 TOR
"Was the bridge constructed in accordance with the design and accepted construction methods and statutory standards applicable at the time?"
(a) The final accepted design for the construction of the bridge is not clear but can be inferred from the undated design calculations prepared by Lt John Armstrong. I understand that alterations to the design were made during construction and these have been added to the Armstrong design calculations. A number of designs were apparently prepared for the project but much of the documentation is now not available.
The letter from the CI SME to Mrs Berryman, the co-owner, dated 2 October 1984 refers to two designs, a footbridge for a concentrated load of 500 lb and a bridge for a quarter ton vehicle. QS sheets covering both bridges were enclosed with the letter but apparently not the drawing. The drawing which was the basis for the QS appears to have been the SME drawing (no number) titled Te Rata Suspension Bridge which shows both bridges. The drawing refers to a design date of Aug 84 and a drawing, tracing, checking and recommendation date of Oct 84. This particular drawing is attached to the Works Consultancy Report dated 5 April 1994 although it has little relevance to the bridge actually constructed.
In a subsequent letter to Mr & Mrs Berryman dated October 1985, the CI SME proposed a new design for the bridge which would allow for a live load consisting of a Landcruiser with a payload of 1 tonne, passengers and or freight. It was to be designed for a 3720 kg load (mass) and on the basis of the use of Macrocarpa timber. It was suggested that Pinus radiata could be substituted for the Macrocarpa. There was also reference to the use of the existing anchorages. The details of the bridge are assumed to be those shown on two SME drawings which are undated and have no drawing numbers. They are both titled Taumarunui Suspension Bridge and numbered 1 of 2 and 2 of 2. Drawing 2 is attached to the Works Consultancy Report also but again has little relevance to the bridge actually constructed. The associated RNZE Quantity Survey Sheets dated 14 November 1985 correspond closely to the drawings but there are some discrepancies and some ommissions.
A comparison of the main features of the various known designs is as follows:
|Item||October 1985||Original Armstrong Calcs||As Built|
|1. Live load (Mass)||3720kg||5000kg||N/C|
|2. Cables||4 no 44mm dia.||2 no 52mm dia.||N/C|
|3. Hangers (Suspenders)||12mm MS rod||13mm SWR 6x19||N/C|
|4. Transoms||300x160 Macrocarpa||300x150 Oregon||2 no 300x75 Oregon|
|5. Stringers (Bearers)||3 no 200x100 2 no 200x50 Macrocarpa||3 no 150x100 Totara||5 no 150x75 Oregon|
|6. Decking||200x100 Macrocarpa||300x75 Oregon||150x75 Totara|
|7. Running/Wearing planks||250x50||Nil||200x50?|
|8. Long Truss||100x25||Nil||Nil|
|9. Towers||Timber||Struc. Steel||Struc. Steel|
It can be seen that significant changes were made to the structure between the Oct/Nov 1985 design and the time of construction. Not only were timber species of lesser durability introduced but also member sizes reduced despite the increase in the live load. Undoubtably the Owners as suppliers of the materials influenced many of these changes.
In summary the bridge was not constructed in accordance with the original Oct/Nov 1985 design and alterations were made, apparently during construction, from the undated Armstrong design calculations. Alterations involving stringers and decking were however recorded in these design calculations.
(b) The concept of structure and construction, apart from the durability of the timber species used, were in my opinion appropriate and applicable. I am not in a position to comment on the construction methods employed at the time in the building of the bridge.
(c) The statutory standards and requirements applicable at the time would have been NZS1900: New Zealand Standard Model Building Bylaw. (Check with SANZ that the bylaw had been adopted by the territorial local authority - Taumarunui or Waimarino County Councils?).
Clause 1.1 Interpretation of NZS1900 Chapter 1: 1985, defines a bridge as a building for the purposes of the Bylaw. Hence there was a requirement to obtain a building permit before commencing to erect the bridge. Almost identical provisions have been adopted in the current Building Act.
Within Chapter 9:1985 of the Bylaw, complying standards were listed covering design, and materials and workmanship. For the project the following standards would have applied:
- Timber. NZS 3603,NZS 3615, NZS 3602.
- Structural Steel. NZS 3404.
- Concrete. NZS 3103, NZS 3109.
There is no evidence to indicate that a building permit was obtained for the bridge nor that the means of compliance standards were followed in the design and construction of the bridge. (It should be noted that at the time, works of this nature for the Crown were exempt from the provisions of NZS 1900.)
GW Butcher 17 Sep 94
TE RATA SUSPENSION BRIDGE
Item 20 TOR
"Where there any additional factors that contributed to the bridge's collapse?"
There are several additional factors which in my opinion contributed to the collapse. The are:
(a) The decision to use two 300x75 beams bolted together for the transoms in place of a solid 300x175 member. The interface was not flashed and permitted the entry of water to the centre of the laminate which, with the oxygen available in the gap, encouraged fungal growth and accelerated the rate of decay. The effective life of the member would have been significantly reduced as a result.
(b) The importance of a regular inspection and maintenance programme for the bridge as a whole and the structural components of the timber deck-structure in particular, does not appear to have been recognised by the owners of the bridge.
GW Butcher 18 Sep 94