In 2008, TxDOT initiated Project 0-6374 to investigate prestress losses in pretensioned concrete girders. The final report from the project included new prestress loss provisions that were found to be simpler, more conservative, and more precise than the current methods outlined within the AASHTO LRFD Bridge Design Specifications. A reference to the report is below:

"Effect of New Prestress Loss Estimates on Pretensioned Concrete Bridge Girder Design"; David Garber, Jose Gallardo, Dean Deschenes, David Dunkman, and Oguzhan Bayrak; Report No. FHWA/TX-12/0-6374-2; October 2012, Rev. June 2013, Published August 2013

This method has been implemented in PGSuper and can be selected via the Project Criteria library Losses tab. The table below is a summary of the loss equations recommended in the report. Refer to the report and the rest of this document for more details.

Treatment of Time Dependency

The TxDOT loss method only computes loss values at jacking, prestress release, and at final. However, losses are required at all defined intermediate stages. The table below lists the fraction of each loss component that is applied in these intermediate stages.

Elastic Shortening and Computation of f_cgp

For elastic shortening at release, the TxDOT 0-6374-2 method assumes a strand stress at release of 0.70 f_pu. The method also assumes that the elastic shortening loss can be computed at mid-girder and is constant along the length of the girder. These assumptions drastically simplify evaluation since iteration is not required, and the losses must only be computed at one location along the girder. The research claims that this method is not only easier to implement, but generates more accurate results than other refined methods.

The equations for elastic shortening along the entire girder are as follows:

Where:

• f_cgp is the concrete stress at the center of gravity of prestressing tendons just after transfer
• e_m is the eccentricity of prestressing at mid-girder
• M_gm is the moment at mid-girder due to girder self weight assuming the span length equal to the girder length
• A_g and I_g are non-composite properties computed at mid-girder

Note that Df_cdp is computed in a similar manner for use in creep loss evaluation.

Assumptions and Limitations of Elastic Shortening Computations - Range of Applicability

The 0.70 f_pu assumption described above is sensible for low-relaxation strands at normal (e.g., 0.75 f_pu) jacking stresses. However, stress-relieved strands and user-input jacking forces can be defined in the bridge model which means that virtually any jacking stress is possible. Also, the research database appears only to consider prismatic precast I-beams with common prestressing configurations: beams with temporary strands, debonding, or arrangements with strands above the beam neutral axis were not considered. In these cases, the 0.7 f_pu assumption could result in incorrect or even negative losses. In order to accommodate for this possibility, and to allow TxDOT engineers to experiment with different methods for computing f_cgp, three different options are provided:

1. Assume that strand stress at release is 0.70 f_pu when computing f_cgp. However, if the jacking stress is not equal to 0.75 f_pu, or the section is non-prismatic (e.g., has end blocks), or if debonded strands are present, or if temporary strands exist: Then use the iterative approach to compute elastic shortening losses as defined in the post-2004 AASHTO specifications. When this occurs, a warning message will be posted to the Status Center.
2. Always assume that strand stress at release is 0.70 f_pu when computing f_cgp. However, if the jacking stress is less than 0.70 f_pu (which would result in negative losses), issue an Error message and halt further computation.
3. Always compute f_cgp using the iterative method, as defined in the post-2004 specifications (LRFD C5.9.5.2.3a).

Time Dependent Loss at Shipping

Note: It is TxDOT policy not to consider shipping during design, however this loss method may be used by other agencies or consultants where shipping is a consideration.

Shipping occurs at an unknown time between release and deck placement and there is no way to compute time-dependent losses for the new method, even if the time of shipping is known accurately. Therefore, users must define the time-dependent loss at shipping using a lump sum value, or a percentage of the total final time-dependent prestress loss. Checking will not be performed to insure that this value is realistic.