In-situ Load Test: a Case Study

Transcription

1 Fédération Internationale du Béton Proceedings of the nd International Congress ID 16-9 Session 16 In-situ Load Test: a Case Study Masetti, F., Galati, N. Center of Infrastructure Engineering Studies, University of Missouri-Rolla, 187 Miner Circle, Engineering Research Laboratory - Rolla, MO U.S.A. Nehil, T. Nehil-Sivak PC, Consulting Structural Engineers S Burdick, Suite 3, Kalamazoo MI 497 U.S.A. Nanni, A. Center of Infrastructure Engineering Studies, University of Missouri-Rolla, 187 Miner Circle, Engineering Research Laboratory - Rolla, MO U.S.A. INTRODUCTION In order to test a structure or portion of a structure for purposes of strength evaluation, two load test protocols are currently available in the United States. The first one is defined in Chapter of the ACI 318 Building Code. It is based on a 4-hour-duration uniformly distributed static load, at a magnitude determined as a ratio of the design load. The second one, still at the proposed level, is based on loading the structure cyclically, in strategic locations, with increasing load magnitudes. This paper provides a procedure for determining patch load/s that applied to the member/structure will generate internal forces (i.e., shear or bending moment), at critical locations, equal to those resulting from the typical uniformly distributed load. The paper describes the application of the proposed load testing procedure for the case of a building scheduled for demolition. The analysis of the experimental results provides professionals with evidence on the validity of in situ assessment based on the proposed protocol for the adequacy of structural members. Keywords: cyclic load, load-test, patch-load, assessment LOAD TEST PROTOCOLS 4-hour-uniformly-distributed load test [1] The 4-hour load test consists in the monotonic loading of the structure up to the designed load level followed by a phase in which the load is sustained for a time period of at least 4 hours. Chapter of the ACI 318- Building Code prescribes that the load must be arranged to maximize the deflections and stresses in the critical regions of the structural elements under investigation, but does not specifically indicate what load distribution should be used. It rather relates the minimum total test load magnitude to the loads adopted during the design phase (Eq. 1): w =.8( 1.4D L) (1) where D and L are the design dead and live loads. Therefore it is possible to apply patch loads by means of hydraulic jacks, and the 4-hour-patch-load-test has been a common practice for many years. Once the structure is conveniently instrumented (where maximum response is expected), initial values of each instrument are to be recorded not more than 1 hour before the application of the first load increment. After the test is started, the load must be applied in not less than four approximately equal increments. If the measurements are not recorded continuously, a set of response readings should be registered at each of the four load increments until the total test load has been reached, and after the test load has been applied on the structure for at least 4 hours. Once the aforementioned readings have been taken, the test load must be removed and a set of final readings must be made 4 hours after the test load is removed.

2 Proceedings of the nd Congress Session 16 Two different set of acceptance criteria are described in order to establish whether the tested member has passed or not the load test. The first one is based on a set of visual parameters: no spalling or crushing of compressed concrete or evidence of excessive deflections that would not meet the safety requirements of the structure are allowed. The second one is based on the measurement of the maximum deflections: they shall satisfy one of the following two equations: lt max h () rmax 4 max (3) where h and l t are the height and the span of the tested member in inches, respectively; max and rmax are the measured maximum and residual deflections in inches, respectively. If Eqs. () or (3) are not satisfied, it is permitted to repeat the load test but not earlier than 7 hour after the removal of the first test load, and the portion of the tested structure in the repeat test shall be considered acceptable if deflection recovery satisfies the condition: rmax f max (4) where f max is the maximum measured deflection after the second test. The 4-hour-uniformly-distributed load test has always been the only method outlined in the ACI building code and has been used for many years because of some of its particular features: successfully holding a test load for 4 hours has a very positive effect on the level of comfort of the practitioner and verifies that the concrete is not stressed too close to its ultimate strength. On the other hand, it presents some drawbacks related to safety (when dead weights are used, there is no feasible way to rapidly remove the load in case of failure) and to the duration of the load application (since there is no guarantee that time-dependent phenomena are experienced over 4 hours). Cyclic Load Test [] The procedure of a cyclic load test consists on the application of patch loads in a quasi-static manner (i.e., sufficiently slow to avoid any strain rate effect) to the structural member, in at least six loading/unloading cycles. Fig.1 reports the cycle characteristics: the load, expressed as a percentage of the maximum applied value is plotted versus the cumulative time. Any load cycle should consist in a sequence of at least five loading steps followed by five unloading steps. For this minimum test protocol, the total load test duration should be about hours with each loading/unloading cycle lasting about minutes. Percentage of maximum test load Load Step{ Load Level A B C D E F Load Cycle Minimum Load Level Fig. 1. Cyclic load test time history Time (min)

3 Proceedings of the nd Congress Session 16 Three distinct acceptance criteria are proposed for the cyclic load test method, namely: repeatability, permanency, and deviation from linearity. The three criteria may be related to any response (e.g., deflection, rotation, strain); however, deflection appears to be the most convenient [4]. Acceptance criteria will be described in terms of deflection in this section. Repeatability is a measure of the similarity of behavior of the member/structure during two twin load cycles (Fig. ) at the same load level, and is calculated according to the following equation: I = Repeatability Index = 1% B B max r R A A max r () P P αref αi P i P max P i Load-Deflection Envelope Cycle A Reference Point Cycle B P ref P min A r B r A max B max Fig.. A couple of twin cycles A B C D E F ref i max Fig. 3. Schematic load-deflection curve for deviation from linearity calculation Repeatability as defined here is not an indicator of the quality of the testing technique, but rather an indicator of structural performance related to recoverable (elastic) deflection and load-deflection response in general. Experience [6] has shown that a repeatability index, I R, of 9% is a satisfactory threshold. For values of I R above this threshold, the member/structure can be considered to pass the load test. Permanency is the relative value of the residual deflection compared to the corresponding maximum deflection during the second of two twin load cycles at the same load level. It should be less than 1% [6] for the member/structure to be considered passing the load test. The permanency index, I P, is computed using the following equation (see, Cycle B in Fig. ): IP = Permanency Index = 1% (6) B r B max If the level of permanency is higher than the aforementioned 1%, it may be an indication that load application has damaged the member/structure and nonlinear effects are taking place. Deviation from Linearity represents the measure of the nonlinear behavior of a member/structure being tested at any time after a given threshold that typically corresponds to its service load level. In order to define deviation from linearity, linearity is defined first as the ratio of the slopes of two secant lines intersecting the load-deflection envelope (see Fig. 3). Fig. 3 shows the schematic load-deflection curves obtained by a total six loading cycles (A through F), that is: three pairs of twin cycles, with each pair at the same load level. The load-deflection envelope is the curve constructed by connecting the points corresponding to only those loads, which are greater than or equal to any previously applied load. As expressed by Eq. 7, the linearity of any point i on the load-deflection envelope is the percent ratio of the slope of the secant line to point i, expressed by α i, to the slope of the reference secant line, expressed by α ref : ( α ) tan i Linearity i = 1% tan ( αref ) (7) The deviation from linearity of any point i on the load-deflection envelope is the complement of the linearity of that point, as given in the following equation: 3

4 Proceedings of the nd Congress Session 16 I DL = Deviation from Linearityi Index = 1% Linearityi (8) Once the level of load corresponding to the reference load has been achieved, deviation from linearity should be monitored continuously until the conclusion of the cyclic load test. Experience [6] has shown that IDL values less than % indicate that the structure has passed the load test. This procedure was used in the case study presented next in order to show its application in details with emphasis on the determination of the equivalent load. CASE STUDY: OBJECTIVE The University Center West building in the University of Missouri-Rolla campus was scheduled for demolition, and, following the cyclic procedure, a one-way RC joist slab (Fig. 4) at the second floor was load tested by means of hydraulics. For this purpose, a portion of the slab consisting of five joists was isolated by means of two saw cuts. This did not change the response of the system (one-way) and allowed using lighter equipment. After the load test, the member was tested to failure. Tested area Fig. 4. Tested one-way joist system When hydraulics are used, the load distribution generally results to be a concentrated or patch-type depending on how the load is transmitted from the jacks to the structure. In the presented case, the structural member under investigation was a one-way joist system, and the design loads (and thus the minimum test load threshold indicated by Eq. 1) were uniformly distributed. The patch load had to be selected in such a way that its magnitude and position caused the same level of internal forces (measured as a percentage of the ultimate capacity) at the same locations as in the uniformly-distributed load test. The next section shows the conceptual steps followed in order to: determine the value of the total test load magnitude, during a preparatory phase; attain the continuous structural assessment, during the load test performance; and, attain the real-time calibration of the test load according to the continuous assessment of the boundary conditions through the measurement of selected structural parameters. PREPARATORY PHASE This paragraph describes the procedure that should be followed by the practitioner performing the load test. 4

5 Proceedings of the nd Congress Session 16 Such procedure was divided in four fundamental steps or phases as described below. Phase 1: Geometry and materials characterization The study of the construction drawings was followed by a field investigation that consisted of visual inspection. Fig. highlights the position of the tested area located at the second floor of the building. From construction drawings: 6" 3" 6" 1" 4" 33" b) Joists assumed reinforcement lay-out After test till failure: 6" 1" 6" 3" 1" a) Plan detail ( 1in =.4mm ) c) Joists real reinforcement lay-out Fig.. Second floor details 33" The portion of the structure under investigation consisted of five joists supported at the end by two spandrel beams. Two transversal joists at the thirds of the span (about 1.ft (3.m)) ensured continuity in the transverse direction. The joists were 31 ft (9.4m) long, spaced in (63mm) on center, and they supported a 3.1in (78.7mm) thick reinforced concrete slab. The joists overall height was 19.6in (497.8cm) with a stem width of in (17mm). The construction drawings indicated that the main reinforcement consisted of 1#9 (8.6mm) longitudinal steel bar for both and positive negative moments (Fig. b). In reality, after the performance of the test till failure, it was discovered that the actual reinforcement consisted of #9 (8.6mm) longitudinal steel bars for positive moment and 1#9 (8.6mm) for negative moment, which was obtained by bending one of the longitudinal bars at the bottom of the joists section (see Fig. c). The slab secondary reinforcement in the transverse direction consisted of # (6.4mm) bars spaced 8in (3mm) center to center. Before performing the load test no experimental data were available and the materials properties were assumed to be those indicated in the construction drawings: the steel yield strength was assumed as fy = 6 ( MPa ) ksi, and the concrete compressive strength as f c = 37 psi (.86 MPa ). Phase : Selection of the type of load test It was decided to conduct the load test using a push-down configuration [6], where the slab above the tested one is used as a reaction to the test load. A single in (17 17 mm) patch load was selected ( in was the width of the joist) located at mid-span of the central joist. Typically the selection of two patch loads at the thirds of the span would be preferable to

6 Proceedings of the nd Congress Session 16 reduce the risk of shear failure. The load was applied by a hydraulic jack connected to a hand-pump, that transferred the load to the slab in the central joist location. As the hydraulic jack extended, it pushed on the steel tube that transferred the load to the test slab, and, on the other side, it pushed on the spreader beam that transferred the load to five joists of the reaction slab. Plywood boards were placed above the spreader beam and underneath the shoring steel frame: in order to avoid any localized damage and to impose the designed load print. Fig.6 shows both a schematic and a photograpf of the test set-up. reaction slab spreader beam plywood steel plates load cell hydraulic jack hollow steel frame steel beam sawcuts plywood sawcut tested slab LVST's transducers Fig. 6. Test set-up Deflections (by means of Linear Variable Displacement Transducers - LVDTs), and load (by means of a load cell) were the only measurements of interest. In order to determine the effective boundary conditions at the span-end locations, for a concentrated load, five displacement measurements were necessary [7]. Therefore LVDT s were placed at the span-end, span-quarter and mid-span locations (along the longitudinal direction). In order to determine the contribution of the adjacent members, it was decided to place LVDT s at each joist stem location (along the transversal direction), at the mid-span location. Fig.7 shows the location of the sensors used in tested area. transversal beam 4" tested part sawcuts 3" load location transversal joists 4" joists LVDT's locations Fig. 7. Sensors position 6

7 Proceedings of the nd Congress Session 16 Phase 3: Assessment by the load-test engineer The procedure that allows defining the ultimate capacity of the structure according to ACI 318- [1] could be summarized into five steps. Following, their description and the numerical values calculated for the examined case will be presented. Since the calculations were performed before starting the load test, the amount of longitudinal reinforcement was that indicated by the construction drawings (Fig. b) (i.e., one bar). Selection of the type of internal forces for the design: in the tested structure, bending moments and shear forces were deemed to be the relevenat parameters. Section capacity for the selected internal forces: Table 1 summarizes the values of the ultimate design forces calculated according to the material properties and geometric characteristics investigated in Phase 1. Tab. 1. Calculated ultimate internal forces Type of internal force Value Positive moment (at mid-span) + φ M = 37kip ft (8.4kN m) Negative moment (at the support) Shear (at a distance d from the support) n - n φ M = 34kip ft (463.7kN m) φ V = 39kip (173.kN) n Rotational stiffness at the span-end locations: it is the determination of the stiffness (namely k s1 and k s ) of the ideal rotational springs at the end sections of the member. They represent the rotational stiffness that is given by the contribution of both the spandrel beams and columns, and the procedure for their determination may follow ACI 318- Sections and [1]. According to the construction drawings and the inspection, the structure resulted to be symmetric with respect to the mid-section, therefore it was calculated k = k = 471 kip ft (6196.kN m). s1 s Determination of the moment distribution along the beam for a uniformly distributed load: based on an elastic analysis, the moment distribution (depending on k s1 and k s ) that results from a uniformly distributed load is determined. Determination of the ultimate design load: it is the determination of the ultimate design load after checking the contribution of all the internal forces. In this case, the uniformly distributed load ( w U ) causing the internal forces to reach the ultimate capacity (particularly shear failure) was calculated as wu = 67 psf (1.78 kn / m ). The maximum design live load ( w L ) was then back-calculated and its values resulted to be wl 17 psf (.1 kn / m ) =. Phase 4: Determination of the magnitude of the equivalent patch load Once the uniformly-distributes test load ( w ) is calculated, the controlling parameters for the determination of the equivalent test load ( w s ), irrespectively of the pattern, are magnitude and location of the load. They depend one from each other, and, once the position has been selected, it is possible to write a general relationship between the two load values []: P = c c L L w= c L L w (9) s 1 l t l t The coefficient c = c 1 c depends on the position selected for the test load, and it accounts for: the fact that the dimension of the patch load in the direction of the main reinforcement is a fraction of the whole member span L l, by means of the coefficient c 1 ; and, the fact that the dimension of the patch load in the direction orthogonal to the one of the main reinforcement is a ratio of the whole member width L t, by means of the coefficient c. The steps needed to estimate an initial theoretical value for the total test load are presented herein. 7

8 Proceedings of the nd Congress Session 16 Determination of the design test-load: it consists in the determination of the uniformly distributed test load according to ACI [1]. For the tested structure, and for the design load derived in Phase 3, the total uniformly distributed test-load resulted to be wtot = psf (11.97 kn / m ), and the maximum load to be applied during the test (the total depurated from the dead) resulted to be w = 17 psf ( 8.14 kn / m ). Evaluation of c 1 : for the selected test set-up, the evaluation of c 1 was determined through linear structural analysis [7] by using the span-end rotational stiffness ( k s1 and k s ) computed in Phase 3. The obtained value was c1 =.71. Evaluation of c : for the adopted test set-up, c was determined using an elastic finite element analysis [7] performed with a commercial software [8]. The obtained value was c =.8. Evaluation of the final load: the evaluation of the total equivalent patch load was obtained by combining the results from the previous steps by means of Eq. (9): w = c c L L w = ft 1in 17 psf = 33kip ( 146.8kN ) (1) s 1 l t Such load is to be considered as an initial tentative value since the actual test load is determined during the load test, based on the measured deflections corresponding to a given load and therefore to the actual boundary conditions. CASE STUDY: LOAD TEST PERFORMANCE Performed load cycles Once all the instruments were connected to the data acquisition system, the continuous recording of deflections and load started. The cyclic load testing procedure proposed by ACI Committee 437 [] and previously described was implemented. First, a preliminary load was applied in order to engage the sensors. Then, after maintaining the load level for about minutes in order to allow the system stabilization, the slab was loaded in six load cycles. The loading procedure consisted in tree pairs of twin cycles (cycles 1,, 3, 4,, and 6) with increasing peak-load up to the maximum test load. Each load step was maintained for approximately two minutes in order to let the structure stabilize under the applied load. Fig. 8 shows the load time-history for the last two cycles only, since the data for the first four load cycles were lost due to equipment problems. 3 3 LOAD (kip) TIME (min) Fig. 8. Last two cycles of the load test 8

9 Proceedings of the nd Congress Session 16 Evaluation through cyclic load test acceptance criteria The acceptance criteria previously described were calculated for the set of cycles 1-6. The test has to be considered failed if one of the three criteria is not satisfied. Table shows that the structure passed the test: only the parameters related to the last two cycles are reported, but all the values were evaluated real-time during the test performance and they were consistent with the results found for the last two cycles. Cycle Repeatability ( 9% ) [ % ] Tab.. Acceptance criteria evaluation Permanency Deviation from Linearity ( [ 1% ) ( [ % ) [ % ] [ % ] Performance Satisfactory Fig. 9 shows, for cycles and 6, the load-deflection curves for different positions along the central joist (midspan and at the quarter-spans) and along the transversal mid-span section (at every joist location). The structure behaves in a quite linear elastic fashion in all curves: it is reflected by the satisfactory performance for all the acceptance criteria. It can be concluded that the structure passed the test; therefore the slab was rated for a total patch load of 3kip (133.4kN ) at mid-span, that produced the maximum level of internal forces correspondent to a uniformly distributed test load w = 1 psf (7.4 kn / m ), that corresponds to a rated live load wl 97psf (4.64 kn / m ) = D D6 D4 3 D1 D1 D7 D8 D4 LOAD (kip) 1 LOAD (kip) D9 D D4 D6 D11 D1 D7 D4 D8 D DEFLECTION (in) DEFLECTION (in) a) Longitudinal direction b) Transverse direction Fig. 9. Load-deflection curves Calibration of the total test load The evaluation of the coefficients c 1 and c (and therefore of the equivalent test load) could have been adjusted through the continuous estimation of the effective boundary conditions by means of the measured structural parameters, that, in the specific case, consisted of deflections measurements at particular locations. For the selected test set-up, the calibration of c 1 was determined through the implementation of the following procedure [7]: 1) the actual deflected shape is measured pointwise by means of five LVDTs: at the span-end, quarterspan and at mid-span locations (D9, D, D4, D6 and D11: see Fig. 7); ) the actual deflected shape is approximated by a mathematical function that fits the measured points: in the studied case both a third and fourth order polynomials were used; 3) once the coefficients of the polynomial functions are known, it is possible to derive span-end moments and the flexural stiffness considered constant along the longitudinal direction; 4) with the informations derived in step 3), by means of linear structural analysis, it is possible to estimate the span-end fixities; ) knowing the boundary conditions in the longitudinal direction (span-end fixities), it is finally possible to recalculate the coefficient c 1. 9

10 Proceedings of the nd Congress Session 16 The calibration of c was determined by using a method based on the application of Betti s theorem [3,4]. Fig. 1 shows the obtained values of the coefficient c = c1 c along with the load time-hystory and the value estimated during the preparatory phase (two estimated values are reported: one was obtained using the longitudinal reinforcement shown in the construction drawings (Fig. b), and the other one was obtained using the real amount of reinforcement discovered after the test to failure (Fig. c)). Since the procedure is very sensitive to data scattering for low load levels, only the calculated values for loads exceeding a fixed threshold have been reported. It has been deemed convenient to choose a threshold value corresponding to the patch load that would cause the theoretical cracking moment at the mid-span section in a fixed-fixed boundary condition situation (.7kip (11.kN ), Fig. 1). Comparing the values obtained from the real deflected shape measurement (continuous line) to those obtained through classical structural analysis methods (dashed lines), which do not rely on measured data, it was observed a maximum relative difference of 13.8% (when the reinforcement shown in Fig.b is considered) and of 3.3% (when the reinforcement shown in Fig.c is considered the real case). COEFFICIENT c (estimated during the preparatory phase considering the reinforcement of Fig. c) COEFFICIENT c (no dimension) calibrated values.6 (estimated during the preparatory phase considering the reinforcement of Fig. b)..1. LOAD (kip) 1 1.7kip Fig. 1. Coefficient c = c c calibration 1 CONCLUSIONS The field test was used to show the logical sequence of calculations to be used in defining a test load for a procedure known as the cyclic load test. Such procedure can be implemented in a computer program and used to perform the real-time testing of a structure. The following conclusions can be derived: Given the uniformly-distributed test load w, it is possible to compute the magnitude of a concentrated load P s based on the geometry and the location of the force of interest. P s is related to w by the coefficient c = c c. 1 1

11 Proceedings of the nd Congress Session 16 c = c1 c can be estimated first by classical analysis methods and then calibrated during the load test performance, until the load is such that the member is linear elastic. For the one-way joist system used as an example, the difference between preliminary and calibrated load values to attain the same maximum internal forces at critical locations was about 3.3%, when considering the actual longitudinal reinforcement. REFERENCES 1. ACI Committee 318,, Building Code Requirements for Structural Concrete and Commentary, (ACI 318-), American Concrete Institute, Farmington Hills, Michigan,.. ACI Committee 437, 3, Strength Evaluation of Existing Concrete Buildings, (ACI 437R-3), American Concrete Institute, Farmington Hills, Michigan,. 3. Betti, E., 187, Il Nuovo Cimento, Series, Vol's 7 and 8, in italian. 4. CIAS,, Guidelines for the Rapid Load Testing of Concrete Structural Members, Concrete Innovation Appraisal Service, CIAS Report -1, ACI International, Farmington Hills, MI.. Lombardo, S. and Mirabella, G., 4, Il Collaudo Tecnico Amministrativo dei Lavori Pubblici, Dario Flaccovio Editore s.r.l., in italian. 6. Mettemeyer, M., 1999, In Situ Rapid Load Testing of Concrete Structures, Master Thesis, Department of Civil Engineering, University of Missouri - Rolla, Rolla, Missouri. 7. Masetti, F.,, Structural Implications of Field Load testing Using Patch-Loads M.S. Thesis, Department of Architecture and Civil Engineering, University of Missouri Rolla, Rolla, MO, USA. 8. SAP Advanced 9..3., 4, Structural Analysis Program, Copyright Computers and Structures, Inc. 11