Basic Design Issues in Timber Frame Engineering THY do we need structural engineering for design of timber Fame buildings? Havent thy stood the tot of time? Well, ye, some bave, and if every vmbet frame building we were planning to construct were based on an exining strctute whore performance was proven, and whose location and loads matched those of the orignal building, certainly there would be litle need for structural engineering in ‘ur designs. However, the sale, complexity and site conditions of ‘many contemporary timber frame projects make taditional ames unreliable for predicting che performance of these new ambitious structures, More and more designs depar from sriclyeaditional forms to find new ways of defining space (Fig. 1). ‘The challenge then i v0 come up with the right dimbers, the sight joinery and the right structural system to ge satisfactory per formance from a design that has never been built before. Structural engineering gives us a consistent method for achieving that goal ‘while nr relying on the mysterious wisdom somesimes ascribed 10 traditional heay timber construction Engineering a timber-famed structure is more than just szing individual joists, beams and rafters. We nced an understanding of how the whole timber Frame and cladding assembly Function an integrated system co support loads. In this aril, the frst of ewo, we will review the structural engineers methiology for building design and look at some ofthe basi strategies we use for accom ‘modating load in timber fame buildings. We will so touch on integrating the design process into construction. We will not get imo’ any ‘higher mathematics here—no equations or number ranching, The emphasis wil be on basic design issues. The Engineering Method. Wh then isthe stucturl engineering approach to design? Once the preliminary size and shape of the is building have been developed, we then proceed through the fol lowing steps 1. Identify and quantify the loads on the structure. 2. Select che member sizes and materials for che structure, 3. Examine how the building behaves under load. 4, Refine materials and member sizes to achieve satisfactory and cffective performance, So, rather chan taking our best guess on an inital sketch and. then proceeding with construction, keeping our fingers crossed, we put our initial assumptions chrough some testing to sce if they are valid and make adjustments as necesary while it sil easy—that is, before the timbers have been ordered. Let’ look a each ofthese points briefly and see how the method works. 1. Identify and Quantify the Load. We have vwo basic categories: loads that weigh dovn on our building asa result of gravity, such as the weight of the structure pls occupants, furnishings, fixtures, and snow aad ice on the tof; and the loads that push sideways or up and down on our building, that is, wind loads or seismic loads. We have to keep in mind that wind passing over and around a stac- ture can also cause suction on the building and create uplift on roof framing For most conventional uses, loadsa building must be designed to support are defined by the building codes. The International Al poor and dewings Tr Nel les cherie ceded Fig. 1. The Herb Nebring Blacksmith Shop at Tillers International, Kalamazoo, Michigan, incorporated large diagonal braces, tension Joinery, diagonally sheathed chearwalls and a flexible roof diaphragm 10 achieve adequate racking resistance for code wind laaas. Desig by Dick Roocenberg, Tom Nebil, Amy Warren Building Code (IBC) is now pretty much our national model code in the Uninad States, Canadian building codes of course apply north ofthe border The IBC prescribes that the typical living areas of a residence need co support a 40 Ibs. per sq. (ps) superim- posed live load; sleeping areas can be designed for 30 psf, attic storage areas for 20 psf. The typical office floor loading require- ‘ment is 50 psf. Commercial spaces used for retail or restaurant use and lobby areas must be designed for 100 psf which is like eov- ‘ering the entire Roor wall 10 wall with sacks of cement stood on end, a prety high demand! For many other applications, such as stadiums, industrial buildings, storage buildings, and the like, the code establishes the requited load capacity. Special applications such as agricultural storage (hay) ot workshop loads such as lumber or timber storage may require you to make your own rational assessment of the maximum likely loads. By “maximum” here we ‘mean the maximum average loading over the Boor area, Stickered hardwood stacked 3. high covering half che Noor would represent 16 TIMBER FRAMING 86 * DECEMBER 2007an average load of about 5) psf (depending how green the wood is and how thick your stickers are). Now, how often might you be stacking that wood 4 Fe high? ‘The building code als> defines snow loads for various geo- gap ars of the couny We wl discus oof loads in more Finally, the code prescribes the lateral loads that our buildings are required to resist. Wind pressures vary considerably with loca- tion and exposure—whether, for example, we are on the rop of « tweeless blu facing the Alantic Ocean in a hurricane-prone region or tucked safely in amongge trees and hills in the relatively placid northern Midwest. Seismic loads similarly vary from location 10 location depending on the likelihood of ground movement. Fortunately for most locations, scismic loads for our relatively ligheweight wood structures do nor control design. Rathes, wind loads are our major challeage for lateral bracing “The building code, by dhe way, not only prescribes the loads we must be able co resis buc also ses limits on how far we can stress the materials we will be wing to build with. A building or stuc- tural element can be expected to perform safely and satisfactorily only if under fall load it is not stressed right 10 the edge of breaking. A safety margin is prescribed so that under fll loading some reserve capacity remains. This helps prevent not only cata- trophic collapse in the event we end up getting that hundred-year snowstorm but also plays 3 part in controling the amount of sag and sway we will seein our construction. The code also ses limits ‘on permissible deflections in floors and roofs and on the sway of buildings under lateral loads. “The building code is often maligned by owners, builders and dlesigners alike, and we have done out share of complaining, but we have to appreciate its role in standardizing che rules of design and construction. By eliminsting guesswork or personal opinion regarding standards of safay and performance—that is, minimum loads, maximum streses and maximum defcctions—the code helps to ensure safety ane dependable performance in consteuc- tion. This is important, not only if you are building for yourself, ‘but especially if you are the consumer purchasing a home or com mercial building from others, We are sure many of you have seen the unhappy results when code requirements are not considered and a combination of bad guesswork, stubborn independence and sometimes downright cheapness produces an unsatisfactory or even unsafe structure. Following the building code, chen isthe first step on che road to taking the guesswork out ofthe design proces. le will generally resule in a safe, conservative design, often sceming over-conservative Ie helps to eemember that the code prescribes or safe performance under fuey severe loading conditions. Following the building code also ensutes a legal design, which chose of us who are registered architects or engincers ae obligated to provide and which you are txpected ¢o provide to obra a building petri. 2. Select Member Sizes and Materia. The initial selections of member size and species for defining a model of a building can be ‘based on some simple rule-of-thamb formulas, or it may be based fon appearance considerations such as how massive (or not) you ‘would like the timbers to xppear in the finished frame. The selee- tions may be based on what’ readily available and affordable. For us in Michigan, white oak and red oak are readily available bus, when I proposed using these non-native species to West Coast clients, they were understandably hesitant. We need to make & pre- liminaty guess at not only species but also grade of lumber, since grade affects the allowable stresses we can use in design and also alfects the code-prescribed modulus of elasticity (elatc stiffness of the material) we use in modeling the structures behavior. Unless the timbers are eo be graded, i’ best to assume the most econom- ical grade reasonable for construction purposes, No. 2, unless you TIMBER FRAMING 86 have appearance requirements that will automatically dictate a higher grade. Selecting higher grades may imply needing to have the timbers graded, however, ensure they meet No. lor Select Seeucrural grading rules. Not al mills are able to provide certified grading, so the expense of hiring a grader could be yours. 3. Examine How the Building Bebaves Under Load. Once we have cstablished the loads for the building and its basic size and shape, and we have preliminary member sizes and species ia place, we see ready to test our ideas. We apply the code loads to = model ofthe structure @ make predictions how the building will behave. In other words, we undertake a structural analysis, “What exactly do we mean by “strcural analysis”? Our models are usually mathematical rather than physical. We idealze the steuccure as essentially a big pile of springs (Fig, 2) Tebow Ne Fig, 2. Structural analysis can be likened to applying lands ta pile of ‘prings—an idealized mathematical model ofthe srueture—and then solving a series of equations to determine the load ie ech spring ‘This type of mathematical analysis is usually refered 10 a stiff ness analysis, since che final calculated loads depend on the relative stiffness of the various components of the model. The sifiness of each member is formulated as a series of equations, a function of the member's size, species, and grade of timber, and can be changed by changing any one of these thee aeibute. I also important in four modeling to consider the stifness of the connections (the joinery) thar relate these members © one another. The structural analysis then amounts to an accounting problem to keep track of which members are pushing or pulling on which other members, hhow hard, and how much they have moved as 2 result of being pushed o pulled. The answers we get from these analyses are the forces thatthe individual members and the joinery need to resist ‘The analysis will also predict for us how much the members will sg, 0F sway: "The complexity ofthe analysis depends upon the complexity of the members and the assembly. Simple elements or Frames can be analyzed quickly by hand whereas complex structures may requite a computer-aided analysis to crank through all the uations. 4. Refine Materials and Member Size. Design 's an icerative process and is certainly easier to make changes to a mathematical ‘model than toa neatly completed frame or building After the first analysis, we examine the loads and stresses on the members and joinery thar have been predicted by the analysis, make changes as necessary to fix those members chat appear co be overstessed, and rerun the analysis to examine how the loads redistribute in the ‘model asa result ofthe changes. This methodology is common to design of structures regardless of the material used for framing. ‘Now chat we've looked ar the basic engineering approach to building design, lers next examine some basic strategies for han- dling loads in timber frame structures. + DECEMBER 2007 7Strategie for Supporting Hloor Loeds. Typical, Hor load is che beginning poine of structural design, and the solution we arrive at will affect our overall building design. We can think of the organi- zation and function of wood framing as being a steady accumla- tion and concentration of gravity loads, somewhat like rain run- ring down through che upturned branches of an imaginary hollow twee, gradually accumulating from rwigs to branches, branches to main limbs and eventually down the trunk to the roors. Similarly ‘our Boorboards deliver the loo loads to the joists, then the joists to the beams, then the beams to the posts, and finally the posts ro the foundations thar rest upon the earth, We can even draw an analogy between the foundations of our building and the roots of| areal tee, which not only distribute the weight ofthe tee to the soil but also help prevent it ftom overturning. Our foundations ‘must perform both these funetions as wel Basics of Foor Framing. Selection ofthe floorboards or roof deck usually does noc cake much effore, We know feom experience that ‘nominal I-in. sheathing boards ae satisfactory to span up 10 2 f and nominal 2-in, tongue-and-groove material can handle spans beeween joists up to 4 fe. The numbers bear us out on this. The bending stresses in decking materials are low under uniform loads. Its the concentrated loads that put the highest demand on any individual boards. Tongue-and-groove joints help to spread those Joads out (provided the tongues actually come into contact with the grooves) so that several boards can participate in resisting a concentrated load such asthe leg ofa pool table o grand piano. ‘Our joist are rypcalysimple-span members: for load-bearing purposes, their ends simply rest on a beam at each end. There is always a temptation co space joists farther apart so that we need fewer of them, and to have them span father, chereby minimizing the number of bents we have to build and reducing the numbec of posts in our oor plan. Sil, we have to keep the bay sizes and spans reasonable to successfully use normally available tieaber sizes, ‘As the spacing beeween joists increases, the load increases pro- portionately. If we need vo span 12 i. between beans in aresiden- tial siuation, we can use 2xs spaced 16 in. on center or 6x8 spaced 48 in. on center. es the same amount of joist macerial cither way, jast distributed diferently. On the other hand, as the joist span increases, the bending force on the joists increases in pro portion to the squate of the span length. In other words, i's no longer a lineae relationship and so we are going to need much stronger and stiffer joists if we want co increase the distance beeween supporting beams. “These same ideas spply to the design of beams. As the spacing bbeeween bents increases, the load on the beams carrying the floor joists increases proportionately. So, ifan 8x10 beam is satisfactory in bents spaced 12 f. on center, chen we may need to go 10.8 9x10 ‘or 10x10 beam to space the bents 14 ft on center. Buc if we want 10 ‘make our bents wider and thereby requir the beams to span farcher from post ro post, the bending fore in che beams will increase as the square of the span. Beams spanning 16 f. have almost ewice as ‘much bending load as do beams spanning 12 fe Fortunately, the geometric section properties that affece che scrength and stfines of our joists and beams are not just propor- tional co the volume of wood used. The strength of a rectangular timber is diecly proportional toi width but proportional t the square ofits depth—and che stiffnes is proportional tothe cube of| the depth. In other words, deeper is better than wide. Inout joist ‘example above, rather than using 6x86 at 48 in, on center, we could also use 4x10s at that spacing. Less lumber ean produce the same strength, and actually beter stifiness. By stiffness, here we mean the ability co resist sagging, and since limiting deflections in our floor framing usually controls our design of long-span flor joists (cemember, the code sets not only minimum requirements on strengch but also maximum limits on sagging), we are especially incrested in the increased stifness that comes with increased depth of the joist or beam. As anocher example, an 8x12 timber laid on the fix is only 50 percent stronger than an 8x8, but the same 8x12 curned vertically is mote than twice a strong asthe 8:8 and more than three times sift ‘Sometimes we need to put openings through ous floor framing that will interrupe the span of our joiss. This commonly happens at stairways, We must remember that the headers and joists chat surround our openings need to be treated like beams: they are col- lectors of increasing amounts of load, a fact thae was often not freed by le ere, al on comm 1 he Foor around stair openings sagging, sometimes alarmingly so, We have to quantify che amount of load on the header and size ic accordingly to handle that load and to span berween joists on either end. Similarly, we need to account for the increased shear and bending loads in the joists on either side of an opening. This dloes not have to be guesswork. “The cutting of notches at the ends of joists and housings in the sides of beams to receive the joists can have significant effects on the strength and stiffness of those members. We will not go into the deals here, but clearly if you cutaway significant amounts of ‘wood from critical highly stressed areas of framing members, you are not going to chd up with the same strength you stared with, Specific limits on notching ate imposed by the INarional Design Specification fr Wood Construction. Working with, oF around, those limitaions isa significant pare of trying to blend timber framing and all-wood joinery with engineered design and modern code requirements. Longer Spans for Floor Framing. What solutions are there for achieving longer spans in timber fame construction? Modern solu tions include the use of manufactured lumber such as LVL (lami- nated veneer lumber, those beams and joists that look like long sticks of plywood) or PSL (parallel stand lumber, referred to by & fiiend of ours not so affectionately as maggot wood) and theit related variations. None ofthese is acceprable for exposed framing, Glued-laminated beams offer a mote visually acceptable soli tion to achieving greater strength and stifiness in timber constuc- tion. Very high quality material canbe used at the top and bottom laminae ofa bear, whetestreses are highest in bending members, and cheaper, lower quality material in the middle of the beam, where stesses are lower. We have used glulam beams in timber frame structures to achieve longer spans and greater load-cartying capacity than could be achieved with solid timbers of greater depth. Not everyone likes the appearance of the glue lines in the timber. With suitable rustication ofthe exposed surfaces and a dark stain, however, ican be dffcule co se che ge lines ifthe laminae are visually consistem, (You will need to order the appropriate appearance grade glulam.) che design calls for curved cimbet, taking a straight imber, sawing it inco flexible strips, bending the strips and gluing them back together v form large curved mem- bers that look like solid timber achieves reslts rliat could not be accomplished by cutting the curve from a single wide stick. (See TF 80, pp. 16-17.) ‘One traditional approach to achieving longer span capabilities with “normal” sized timbers isthe use of so-called keyed beams ig. 3). Pofef=f=p Fig, 3. Schematic drawing ofa keyed beam. FEF] [we take two timbers and sack them one on top ofthe othe clamp them cogether with bolts and insert mechanical "sip 18 TIMBER FRAMING 86 + DECEMBER 2007resisters” co prevent the upper and lower members feom slipping past one another when loaded in bending, we can make the two timbers behave almost aif we ad one sold picce of imber of the combined cross-section. There is some tendency for slip to occur along the mating surfaces and so the keyed beam is nor quite as stiff asa solid piece of timber would be. There is also a great deal of work in cutting aad assembling keyed beams. They have been used for hundreds of years and provide us with a means of fabri- cating timber much bigger than could be harvested fiom readily sxailable tres. So, for example, cwo 8x10 beams stacked and prop- erly keyed can be neatly as effective as an 8x20 timber, chat is about twice as strong and three times as stiff asthe unkeyed 810s simply stacked. “Another traditional way 10 achieve long spans in floot framing ‘with noemal timber sizes is by erussing. Timer wusses can readily be buile wo span 30 co 60 f., and longer spans are possible. Trusses used for oor framing will eypically have both the top and botcom members pazllel and horizontal, so the teuss outline becomes 2 rectangle rather than a tangle such as is used for eoof framing, ‘They often incorporate some metal, for cxample steel rods for ten- sion members, oF straps and bots used for eeinforceent at highly stresed joinery. Tn overall structural behavior, most sruses act like deep beams Sinpy spanaing from support fo support. Their depeh, however, commonly in the range of one-sixth to one-tenth of the span length is one of thee disadvantages. 1f you plan on spanning 30 fe across large oom, you will probably need a cuss about 5 Fe. deep. Design and cutting ofthe of-required serious tension joinery can be complicated. Trusses are expensive and labor intensive to design and build, bu chir appearance can bea significant aesthetic plus to a timber fame building. Ifthe depth of erusses at intermediate floors within a building is {impractical —think of adding 5 ft. o the rise of ase of stats from fist to second floors—then we can always position the crusses up a the 200f and then hang che intermediate Boors from: the trusses boy means of hanger rods (Fig. 4) install erusses a full tory tall in the interior pation walls of upper-level floors, frame che Roor joists co the trusses and leave the floor below clear of poss. This strategy likewise dares back at least to the 1800s, Inserior Suppors for Floor Framing. As the width of our timber frame buildings increases, clearspan framing is often impractical ‘We re mot likly to use intermediate supports o reduce the spans of our floor feaming and thus avoid the need for excessively large timbers or the more complex solutions discussed above. If the second-floor poss in the design line up over the first-floor poss then the posts can be continuous from top to bottom, which lets us collet the loads from the bears and direce them immediatly (0 the foundation chrough the posts. Ofien, however, differences in layout and function berween first and second floors require that interior second-fioar pests be offset from chose onthe ist oor. In these cases, we rely on the second-look beams to collect the loads from the second-floor posts and transmit them eo the fsu-floor Posts. Since offset posts can impos large bending and shea loads in the second-floor beams, these beams need t0 be specifically dlesgned for the amount of load and the spans involved if we are going o avoid excessive sagging of the second-floor beams (Fig. 5) Fig. 4. Hanging the second floor from trusses in the attic or roof can leave the frst flor five of posts. ‘This solution was commonly used during the 1800s in indus- ial buildings and especially in cheatets, where the balconies were ung from large timber trusses in the roof framing. We can also ‘TIMBER FRAMING 86 Fig. 5. Cros sction of 100-earold aati fame barn in southern “Michigan. Offit poss can case ageing ofrupporting bears nls pst Toad and beam bending stress are adequately accounted fr Cantlevering exteroe walls beyond fs lor walls sin gazrison- style construction i a special example of offer poss. Here agai, second-floor beams need to be specifically sized for loads imposed fn their outside ends to avoid a sag outside the outer walls and a hump inthe interior main span ofthe cailevered bear. But can- silvers can thus be used ¢o advantage to eduace maximum bending seres in the main span, with the cantilevered portion acting asa lever tieving main span sig Bolsters (capitals) can be used ro increas the effective width of postsand thereby decrease he elective span of beams. The bolsters serve then co stiffen the loots and inereate the shear capacity of the for faming in the vicinity ofthe poss, Since they inerrupe the posts atthe oor, ther use is restricted to buildings with lvel-by Kevel construction, such as I9eh-century industial buildings Bolsters can be useful for support of the first-floor beams in a simber frame structure, installed over the posts in the lower level. Knee braces provide us with another means of efecively increasing the width of our suppore and reducing beam spans + DECEMBER 2007 19Fig. 6. Knee braces can effectively widen a support and stiffen beams, belping to transfer loads to posts. Knee braces can reach out farther than bolsters and provide sifer support for beams, The longer they ar, the beste job they can do. They also offer some design flexibility ae beam-to-pose joints because posts can run through vertically or beams can run through horizoneally (Fig. 6) When a knee brace is loaded by a beam, it transmits that load to the post in a direction parallel to the brace axis, The brace not ‘only pushes downward on che post bur also sideways (and with ‘equal force in a 45-degree brace). At an interior post where there may be knee braces on both sides, chere is no net sideways thrust fon the post when loads are balanced on the spans above. At an exterior post, however, there is no balancing load coming in from the outside, and so the post is subjected to a combination of axial loading from the weight of framing above and bending loads from the sideways thrust of the brace. Knee braces act like wood arches, and there is always ourward thrust in an arch. The knee braces are trying co push the posts outward, and that means we need to have both adequate bending serength in our posts and restraine at the top and bottom of the posts to keep them from being forced out- ward. Thus we need co check the tension capacity of the beam-to- pos joinery above the knee brace and the shear resistance at the base of the posts (Fig. 7). Fig. 7. Behavior of simple knee-braced frame under gravity load. (One last paint: how do we make a “nonstructural” knee brace? ‘That is, if we invend to insall knee braces in a frame just for appearances, i it possible te prevent them feom putting sideways laads on posts when beams above deflect under loading? We come across a lot of nonstructural knee braces in old barns where the braces are rating around in their housings or have even fallen to the floor below. Shrinkage of the post and beam can inwoduce some play in a knee brace that originally fe perfectly; the beam has to sag somewhat to close the gap. That is ypieally an unintentional effect If we really want knee braces to be nonstructural, we need to provide gaps atthe bearing surfaces in housings and use small- diamecer flexible pegs. Ignoring the posible inadvertent loading of a nonstructural knee brace on a post may not be of any conse- quence at shore stubby posts and beams, but could produce some Undesired effects in al, more flexible frames. Remember that exte- rior posts in a building may already have a significant bending load ‘on them fiom wind suction. Strategies for Handling Roof Loads. Roof loads consist of the ‘weight of the roof framing and roofing materials plus any super- ‘imposed live loads in the form of snow, ice, ot maintenance per- sonnel and equipment. Snow loads ace highly variable in many regions of the country. In Michigan and upstate New York, for example, lake-ffect conditions cause increased snow-load requite- ments for sites close 10 the Great Lakes. In mountainous areas, snow load requirements vary with elevation and exposure. For design purposes, i's mandatory that you contact the building off- cial where a new project is to be constructed to obtain the local snow-load ordinance. Tr nor that roof loads ae diferent ftom floor loads; both are ‘xavig’ loads pulling down on the frame. Is just that we usualy choose 1 frame roof with some significant slope. If (heaven forbid) all roofs were fat, then our stategies for supporting roof loads would be identical co those for oor loads, but sloping roofs cdo a much better job of shedding rainwater. Ifthe rook is steep ‘enough and slippery enough, ic can shed snow loads as well. Pairs ‘of opposed raters, however, can produce a significant additional effect they may generate outward thrust on the walls. ‘Shed Rafirs (Fig, 8). Just because a rafter slopes does not x0 matically imply that ie will generate ehrust. A sloping rater sup- pporced at both ends by walls or beams, as in a shed roof, generates no thrust at all Ie aces the same as a perfectly at oor joist. Why is chat? Wall, both ends are simply supported to prevent the rafier fiom moving downward, and there are no laetal loads om this rafts just the weight of snow and gravity drawing ie seraighe down. So, the reaction a each end ofthe rafters sraight up. I's when We put rafter pairs together opposing one another that intersting things can happen, Ridge-supported Rafters (Fig. 9). Ws possible ro have rafter pars thar behave like shed rafters. Thats what we ge when we incorpo- rate a structural ridge beam into the framing. The ridge beam (posted or otherwise supported) prevents the upper ends of rafters from moving downward under the weight of eof loads. The lower ‘ends ofthe rafters ae similarly supported on either beams or walls and so cannot move downward. The net effect includes no ourward thrust. Steucraral ridge beams do pose some challenges for us. They carry haf the rafter span loads on cither side and 50 can support significant areas of oof. They span between distant supports such as ‘russes, interior king posts oF gable end walls and can very quickly become large heavy timbers oF even trusses if ee distance berween supports becomes large. Steucrural ridges must be designed just ike floor beams to resist bending and shear forces. Pualin-supported Rafters (Fig. 10). Instead of making the ridge beam do al the work, we can split it up into 2 pair of supports located at some faction of the length along the rafer span. We 20 TIMBER FRAMING 86 + DECEMBER 2007Fig. 8. Shed rafters, supported at each end, produce no thrust. row have «wo beams (again, themselves supported) helping «0 carry the load of the roof instead of just ane ridge beam, and $0 we may be able to span farcher or use lighter beams. These i no hard rule where che purlin beams should be located along the rafter span. In practice, because of functional requirements inside the building, we find char purlins lie most often in the range of one half ro two-thirds ofthe distance from plate to ridge. Provided the paslins ate more chan halfway up the length ofthe rafters and the {afters are uninterrupted from plate to tdge, the pusins do a firly good job of supporting the upper ends ofthe rafters and reducing hes from the rafter pars. In other words, che rafters ina purin= supported roof can sil behave almost like rdge-supported rafter pairs. The farther the puslins are located fiom the tidge and the lighter and more fexile che rafters, however, the more cheust will be generated by the upper span ofthe rafters under load. Ler’ move on to look at what happens when we remove the ridge beam and the pucin beams completely from an opposing rafer pat. There is nothing to support the upper end of the rafters and, as the ridge line moves downvvaed under load, the geometry Of the sloping rafters causes thee lower ends ro move ourward, thereby generating horizoneal thrust. If we do not want our ridge to come down, and thae is usualy the goal, we need to restrain the luter ends of the rafces from moving apart. Thar can be most directly done by some stuctural element that “pushes back” to pre- veri the spreading of the rafter feet. The resisting element could take the form of massive masonry walls, simply too heavy or too well but- tresed to be tipped over by the thrust imparted by che rafters. The ‘external resraine does no have co be masonry. It can be provided by the intersecting walls of other portions ofthe building framed per pendicular to che dizection of thrust. Basetied rafies (Fig, 11). Rather than extemal restraint, we often use some form of internal restaint. The most efficient approach isto tie the rafter fee wget with an internal tension tmomiber The ie cat be timber, it an be» ee od oi cin be dimension lumber. Placing the tension tie tight atthe foot ofthe ers, at the level of the plats at the top of the supporting walls, pts the restraint right atthe point of application of the chrust and creates a basie truss, The rafters now act somewhat like an arc: they carry axial load from the ridge down tothe pate, but they also have to resist bending berween the ridge and plate caused by roof dead load and snow load. Colla sues can help brace the rafters and reduce sagging; well gee to that in the discussion of cusses that follows shorly Base-tied rafters are more complicated than shed rafters or tidge-supported rafters, which sustain no axial loads. Properly designed rafters tied atthe plate take into account this combina- tion of raial loads and bending loads. On steep pitches, say 12:12 Fig. 9. Ridge-supported rafters act lke individual shed rafters: Fig. 10. Purlin-supported rafters produce lee thrust at plate Fig. 11, Basetied rafters directly solve thrust but sustain axial loads. or greater, the effect is negligible. As the slope gets low, ia can be significant and the raters may need to be heavier chan simple span ralters otherwise would be for the same span and roof loads. For 2 6:12 slope, the axial load in the rafter is a litele over evice che gravity load reaction ae the support, whereas for a 4:12 slope ie is fver three times the reaction and at a 3:12 slope more ¢han four times the reaction. Puc in a summary way, at 12:12 the horizontal reaction is balfthe gravity load, at 6:12 our-chrust equals vertical load and, at lower pitches, the theust exceeds the gravity load. TIMBER FRAMING 86 + DECEMBER 2007 arait Ties (Fig. 12). 1s not uncommon for a designer to want to move that tension tie at che plate up “just a ltde bit” 0 gain increased headroom without having to increase the height of the building. Such a move can also allow an interesting vaulted ceiling effect. As we raise the tension tie up along the rafter span, however, some additional challenges begin to develop. Ths discussion is per= «inent not only to simple rafter pairs such as re found in common- rafier roof assemblies, but also to principal rafter pats oF cusses ‘when the bottom chord (Cension ti) does not intersect the top chord (rafters) at the supporting beam or wall. The rafter paie stil is subjeced to bending loads applied berween the ridge and outer support under the action of roof loads, but in this case the raised ticadds to the bending loads by pulling inward a its connection to the racer Raised ties ae an unsatisfactory way co resist the thrust of rafters because they pur such large bending load into the rafters and alo senerate large forces to be resisted a the joinery between the tie and he rafters. Furthermore, joinery from tie to rafter weakens the rafter right at its most highly stressed location. ‘The higher the tie is above the plate, the tougher these design challenges become and the heavier the rafters need to be to avoid excessive sg in the roof and ourward bulging ofthe walls Dropped Ties (ig, 13). Instead of raising the tension tie above the level ofthe plate, I9eh-century American framers dropped it below the plate, for example in the high-posted (story-and-a-half) capes of New England and many baras throughout the country. The dropped tic gready simplifed the joinery required by the English tying joint, which it largely displaced, by bringing the tie bbeam in below the poine where post, plate and principal rafer would meet ‘There is a trade-off inthis system, though. Similar to che raised tie, the dropped tie removes the restraining element from the point of application of the crust, the Fee of the rafters. The eesuleis that ‘we induce bending inthe posts rather than in the rafters as was the case withthe raed tie. As the distance from plate to tie beam level increases, so do the tension in the tie beam and the bending loads in the poss. Here again, joinery from tie to post weakens the post at a highly stressed location, In barn framing in the Midwest, a common rule was never to drop the tie more than 2 fi. ftom the top of the plate. Even this guideline was not enough to prevent _many tension joinery failures at the joine berween tie beam and post. The problem of tension loads at a dropped tie beam-to-post joine is further exacerbated by the presence of knee braces and wind loads So, placing che cension tie at the level of the plate isthe most efficient way of resisting rafter thrust, That does not mean the other options are not available to us, just that we need to design forthe extra bending and joinery forces involved. Truser (Figs. 14-18). Similar to the challenges we face with increasing spans in our floor framing, we face limits on what we can do with the simple triangular truss represented by the rafter pair with the tension tie at the plate. As the span beoween sup- porting walls or beams gets large, the bending loads and thus the sag in the rafers and in the tension tie increases—particulariy when the tie supports a ceiling or attic storage space. This forces us to go 10 ever larger timber sizes until that simply is no longer a practical or economical solution, Buc justas we solved this problem in our floor frarsing by introducing intermediate supports along the length ofthe span, we can do it in our roof framing aswell, not by putting more posts in che buildings o suppore the roof framing {although that works too), bur rather by installing internal sup- ports within the roof framing assembly. If we place a kingpost in the simple triangular tuss (Fig. 14) we can greatly reduce che sug in the tension tie. The kingpost acts as a hanger and essentially pulls the gravity loads up from the vie beam Fig. 12. Raised tie introduces additional bending loads in the rafters ‘and inereace tension inthe te Fig. 13. Dropped tie introduces bending in the pots and increases ten- ion inthe tie. Fig 14, Kingpost reduces sag (and thus tension) in te beam of a base tied truss. Rafter sag can be soloed by struts fom kingpost to rafter. 22 ‘TIMBER FRAMING 86 * DECEMBER 2007ico the rafters, thereby increasing the thrust on the rafters and increasing the tension load ar the joine berween tension tie beam and raters. Thar helps the tie beam, bue what about the rafters? We can insert an intermediate support there as well in the form of a horizontal colar strut that prevents the slers from sagging invaed at midspan (Fig. 15). Fig. 15. Invoduction of compression collar beroees the rafters lfens them against sag. ‘This solution also increases tension in che joinery beeween tie beam and rafter buc greatly stiffens the rafters and allows us 10 reduce cheir heft we pur these evo ideas together, we have an early form of roof truss found in European churches of the fourth and fifth cencuries (Fig, 16). A modern configuration is seen in Fig. 17, Fig. 16. Together the kingpost and the collar struts form a 4eh-century Roman ernst. Hammer Beams. No discussion of roof Faring in tember build- {ngs would be complete without at least a brief look at hammer- beam roofs People often mistakenly look at hammer-beam roof framing asa means of pulling yourself up by your boorstraps—that is, somehow achieving lage cleat spans without a tension tie and yet avoiding the problem of thrust from the rafes. le just aint 50 Rather, think of the entire hammer-beam assembly as forming simply a large rafter pair with no support a the ridge (Fig. 18) Fig. 17. Typical modern kingpost truss with diagonal web members to support rafters. White oak 6x12 rafters and tie beans spar 30 fe Fig, 18, Hammer-beam trues oan be understood as pairs of rusted ‘afters thrusting outward againee ther restraints. ‘The actual behavior is somewhat more complicated than that, and these seeuetures are properly called hammer-beam trusses. (See ‘TF 48,) Sill, oof thrust comes down through the lowest diagonal framing member to che supporting wall or timber post, and that thrust is going co have to be resisted or the ridge will come down and the supporting walls or posts will move outward. If we do not hhave massive masonry walls with external buttresses or walls from other parts of the building functioning as external restraint, then wwe will need ineernal tension ties or hefty post and some serious tension and compression joinery atthe intersection of the post with the lowest diagonal brace and with che rafter. We also will likely need restraint against outward movement atthe base of the post. ‘There is no cookbook formula for these forces and the associated joinery. These depend on che spans, the pitch and spacing of the trusses and the height of the walls, and they must be specifically engineered if satisfactory performance is o be assured. Tow Nett and Avy Warnes’ Tom Nebil (tnchil@nchilsivak.com) is a principal at NebileSivak Consulting Structural Engineers in Kalamazoo, Michigan. Amy Warren isa seructural engineer at Nehil-Sivak. The second part of his wo-part article will discuss lateral loads and the proper use of struc ural engineering in timber-feamed buildings. TIMBER FRAMING 86 * DECEMBER 2007 23