ChangeLog.txt

			WT_Perf Change Log
			------------------
Ancient history:

DarkPast - PROP created by Aerovironment Inc.
04/01/85 - PROPPC ported to the PC by J. David Kocurek of Computational
	    Methodology Associates.
Later    - Modified for use by SERI/NREL by Jim Tangler.
1994     - Ported to SGI and slightly improved by Marshall Buhl of
	    NREL-NWTC.


1.00  25-Apr-1996  M. Buhl
 Modified for variable speed.  Many utility routines were stolen from
      some of Buhl's other programs.  All propeller code was removed.
 Renamed to WT_Perf.  Input file was reorganized.

1.10  29-Apr-1996  M. Buhl
 Modified to have which variables should be output as an input parameter.

1.20  10-May-1996   M. Buhl
 More clean-up.
 Fixed old bug where tip loss was not calculated if the hub radius was zero.
 Improved calculation of local wind shear by including the effect of shaft tilt
      and precone.
 Fixed old bug where local Q was calculated using tip-speed ratio instead of
      local speed ratio.
 Added ability to input metric and/or dimensional data.
 Rederived the equations for the rotor coefficients and dimensional data and
      changed the algorithms.
 Added date and time stamping of output files.

1.30  13-May-1996   M. Buhl
 More clean-up.
 I fixed up the metric and dimensional versions.
 I renamed the ratio variable to not have the ratio in them (e.g., I renamed
      HH_RATIO to HUB_HT).

1.40  14-May-1996   M. Buhl
 I rewrote the post-stall routines.
 I moved many one-time calculations to a new routine (SETUP_V) and stored them
      in a new common block and created the new include file (viterna.inc).
 I modified the Viterna calculations so they use the stall Cl and Cd instead of
      the last ones in the tables--this was an old bug.
 I cleaned up a few minor things.

1.50  14-May-1996   M. Buhl
 Added option for tab-delimited and fixed-width output.

1.51  15-May-1996   M. Buhl
 Fixed some minor bugs.

1.52  30-May-1996   M. Buhl
 Fixed a bug in setting the minimum/maximum number of sectors.

1.60  07-Jun-1996   M. Buhl
 I reordered the input file.

1.61  10-Jun-1996   M. Buhl
 I broke out the system-specific routines to sys_*.f for easy porting.
 I cleaned up some minor problems to do with porting to LF90.
 I fixed it so there are no newlines between segments of blade element data if
      there is only one section.
 I renamed STRLEN to STR_LEN and moved it  into the sys_*.f files.  For LF90, I
      use it to call LEN_TRIM.
 I moved GETARG to the sys_*.f files and renamed it GET_ARG.  I had to create
      one using GETCL for LF90.  I also needed to create a PARSE routine to
      break the command line into tokens (words).  I put it in sys_pcl.f.

2.00  17-Jun-1996   M. Buhl
 I added the ECN 3D effects for Cl and Cd.
 I moved MAX_ITER from a compiled parameter to an input variable.
 I fixed it so that when we fail to converge, we actually see -999.99 instead of
      some other number.
 I added one digit of precision to the dimensional output variables.
 I created a new program PROP2WTP that converts PROP files to WT_Perf files.  It
      uses the same include files and also the gen_use and sys_* files as
      WT_Perf.  This should make it easier for folks to use this new program.

2.01  03-Oct-1996   M. Buhl
 I changed the FORMAT statement used to output I_SHFT so it was compatible with
      integers.
 I put in a test to make sure N_SEG doesn't exceed MAX_SEG (Thanks for finding
      that one, Palmer!).  I added

2.02  26-Nov-1996   M. Buhl
 Palmer Carlin discovered that for high tip speed ratios and large pitch angles,
      the turbine acts like a fan and the angle of attack becomes a fairly large
      negative number.  If the lift and drag tables do not cover the potential
      range, the interpolation routine would return silly values. I changed
      GET_COEF so that it prints an error message and stops.

2.03  08-May-1997   M. Buhl
 I eliminated the test for alpha being less than alpha_stall for the loss
      calculations.
 I eliminated the dummy argument N_SEG from PS_LIFT and PS_DRAG.
 I modified PRAND so that I don't waste time on calculations when the losses
      will be very small.
 I simplified the calculation of the advanced brake model.

2.04  30-May-1997   M. Buhl
 I decided that the thrust and torque per unit length for the blade element data
      should not be multiplied by the number of blades.  It is now output as per
      blade.

2.05  17-Nov-1997   M. Buhl
 I changed the date format to use dd-mmm-ccyy.
 I also changed a lot of stuff to use F90 intrinsics.
 I ported to the DEC Visual Fortran compiler.
 Peter Schaffarczyk helped me find a way to make the code friendlier. If one
      entered a hub height so small the blade struck the ground, then the shear
      equations would blow up.  I put in a check to make sure the hub was high
      enough during the input phase.

2.06  06-Feb-1998   M. Buhl
 I ported WT_Perf to Sun Fortran 90 and may a few minor changes.

2.07  20-Aug-1998   M. Buhl
 I changed the axial induction-factor calculation to eliminate applying the
      tip-loss correction factor twice.  The new equation is 4aF(1-a)=Ch instead
      of 4aF(1-aF)=Ch.
 I also modified the use of Glauert's empirical formula so that the coefficients
      of the quadratic are now a function of the losses.
 I also eliminated the IND_FACT routine and simplified the calculation of both
      the inducation factors (IF) so that the new IF is the average of the old
      and the proposed IFs.
 I also changed the IF iteration tolerance to 1e-7 and changed it so both a and
      a' use the same tolerance.  The old version made the a' 10 times harder to
      converge.
 I added the program name,  version, and version date to the blade element
      output files.

2.08  16-Dec-1998   M. Buhl
 I modified INT2LSTR to compute the length of the resulting string.

2.10	23-Jun-1999  M. Buhl
 I modified the induction-factor calculations so that they used the drag terms.
      This is how BLADED does it.

2.11	07-Oct-1999  M. Buhl
 I modified the output format for the Cp data when generating column-delimited
      files so that the numbers line up correctly.

2.12	08-Oct-1999  M. Buhl
 I modified the induction convergence so it uses a method similar to AeroDyn to
      prevent jumping over the solution.

2.20	25-Oct-2000  M. Buhl
 I removed the drag terms from the induction equations that I had added in
      v2.10.
 I cleaned out some alpha-code left over from the code runoff.

2.21	25-Jan-2001  M. Buhl
 Jason Jonkman found that the pitch values listed in the output file did not
      agree with that in the input file.  For some unknown reason, I was adding
      the tip twist to the pitch value for the output.

3.00a-mlb  17-Nov-2003  M. Buhl
 I rewrote the code in Fortran 95.
 I made some changes to the input file and disabled all the 3D stuff.
 I fixed it so it used the standard NWTC routines.  I made very few changes to
      the algorithms.  I got identical answers for the original test case.

3.00b-mlb  18-Mar-2004  M. Buhl
 I eliminated the old style of including airfoil tables in the main input file
      and changed it so it uses AeroDyn input files.
 I also created a new style for the airfoil tables and the code reads both
      styles.  The airfoil-reading routines were put into NWTC_Subs.
 I added Reynolds number interpolation.
 I reordered the main input file some.
 I modified the induction equations so that the user can choose whether to
      include the drag terms or not.
 I added the option to decide whether to use the old PROP induction or
      the newer PROPX/AeroDyn algorithm.  I added the PROPX algorithm.
 I moved the output sections of the input file to the end of the file.
 I fixed the flap-load equation to compute it at the hub radius and to account
      for the fact that rotor could be preconed.
 I replaced the equations to calculate the element Ct, Cp, and Cq with equations
      Pat Moriarty gave me.  It improved the correlation to the rotor Cp.
 I added the total losses to the element file.

3.00c-mlb  22-Apr-2004  M. Buhl
 I added the variable "Yaw" to the input file and modified
      WTP_Subs.f90\GetData() to read it.  I modified the calculation for the
      minimum number of sectors to also check for yaw error.  I modified the
      VWndNorm and VWndTang calculations to account for yaw.
 I created new non-dimensional variables for hub radius and hub height so that
      it is clear in the code which is being used at any given time and to avoid
      repetetive multiplications.
 I rederived the equations for the normal and tangential wind speed and used
      them in WTP_Subs.f90\GetVel().

3.00d-mlb  22-Apr-2004  M. Buhl
 I changed the blade-element data to mimic what we plan to make AeroDyn look
      like in the future and allowed variably spaced elements.
 I modified it so that only unique airfoil tables are read in as is done in
      AeroDyn.

3.00e-mlb  12-May-2004  M. Buhl
 I changed all occurances of Rn to Re to use the most-common notation for
      Reynolds number.
 I eliminated most calling arguments on WTP_Subs.f90\GetVel() because they
      conflicted with the ones in the WTP_Data() module.  I retained only
      VWndNorm.
 I got rid of the old speed stepping logic because it had been replaced long ago
      by the DO counters.  It was sometimes causing the last step to be skipped.

3.00f-mlb  15-Jun-2004  M. Buhl
 I added the new combined-case parameter-specification format.  I moved almost
      all of the analysis and I/O from the main program to subroutines in
      WTP_Subs.f90.  They are CombCase, ParmAnal (the old method), and RotAnal,
      which is the inner part of the analysis (covers all sectors and segments).
       Many variables were moved from the main program to WTP_Mods.  I had to
      make changes to the output section of the input file.
 I added the new KFact flag to the output section of the input file.  This flag
      allows the users to say if (s)he wants output in K.  For example, thrust
      would be in kN instead of N.
 I added an extra digit of precision to the output.

3.10a-mlb  22-Jun-2004  M. Buhl
 I added Cq as an output parameter for combined-case analyses.

3.10b-mlb  02-Jul-2004  M. Buhl
 I modified the code to work with the new version of the NWTC library
      (v1.00a-mlb).

3.01.01 24-Apr-2006  M. Buhl
 I converted the program to use the new NWTC Library.
   -- I eliminated the Pi constants from WTP_Mods.f90\Parameters.
   -- I replace all USE statements with USE NWTC_Library.
   -- I changed the calls to the UserTime function to use the intrinsic
      CPU_TIME() subroutine.
   -- I replace the variable Rho with AirDens because that was done in the
      library.

3.01.02a-mlb 24-July-2006  M. Buhl
 Paul Kreiner found a case that caused the angle of attack to become NaN, which
      then resulted in an error in the coefficient lookup.  The problem was not
      easily tracked, as it did not occur when run in Debug mode or when some of
      the pertinent variables were printed to the screen.  I made the problem go
      away by limiting the proposed new value of the induction to the range
      [-1:5].  I added one line to WTP_Subs.f90\Ind_Prop() to fix it.

3.02.00a-mlb 03-August-2006  M. Buhl
 Rewrote Ind_Prop to use Newton-Raphson for the iteration.
 Replaced the parameter, Epsilon, with the intrinsic function, Epsilon().
 Replaced the IndProp input flag with the IndType switch to enable turning off
      all induction calculations.

3.02.00b-mlb 22-August-2006  M. Buhl
 I completely redid how Ind_Prop works.  It now tries to find a zero crossing
      for the error function before trying Newton-Raphson.
 I eliminated the PROPX and Classic Brake State options, as I see little use for
      them.  This included eliminating the third option (=2) from IndType and
      the AdvBrake flag from the input file.

3.02.00c-mlb 24-August-2006  M. Buhl
 I added the OutNines flag to the input file.  If set, unconverged solutions
      will output nines.  If not, the final attempt at an induction solution
      will be used for performance calculations.  If no solution is found,
      which is different than being fully converged to satisfy the tolerances,
      nines will always be generated.  I added the total number of cases to the
      warning message.
 I added an input option (Beep) to enable/disable beeps on errors.  It is also
      used in the NWTC Library.

3.02.00d-mlb 28-August-2006  M. Buhl
 I calculated a' as a function of a and replaced the 2D Newton-Raphson routine
      with a 1D routine.

3.02.00e-mlb 30-August-2006  M. Buhl
 I tried to get the a 1D iteration on phi, but there are too many zeros.  I
      ignored the Glauert correction.

3.02.00f-mlb 31-August-2006  M. Buhl
 I added the Glauert correction and used the correct equation for a'.  It didn't
      help.  I give up with this technique.

3.02.00g-mlb 31-August-2006  M. Buhl
 Reverting back to v3.02.00c-mlb.
 I renamed Ind_Prop to InductBEM.
 I added a binary search for when the Newton-Raphson fails.

3.02.00h-mlb 13-Sept-2006  M. Buhl
 I moved GetAero outside of InductBEM.
 I had to decrease the step size in the zero search to avoid a situation where
      the error function just barely crosses zero before returning.  I found
      cases where it jumped right over the lowest AI solution.  This
      dramatically slowed the program (maybe a factor of 10), but made it much
      more robust.  WT_Perf now seems to find solutions for all the CertTest
      cases and for the one Paul had problems with.  Unfortunately, CertTest
      Cases 3 and 5 get different answers, but the answers do seem valid.  I
      guess the two codes just converged on different solutions.

3.02.00h-mlb 13-Sept-2006  M. Buhl
 I decided to test for change of sign of slope in addition to change of sign in
      the zero search.  The logic is that if the slope changes sign and the
      error function does not, a minima or maxima has be passed.  If that minima
      or maxima just kissed the zero line, coarse steps could jump over it.  A
      non-zero-crossing minima/maxima will now be searched at a fine resolution
      and I can decrease the resolution for the normal steps.

3.02.00k04-dcm 22-June-2009 D. Maniaci
 Added file output for plotting ClBem (Cl to give AxIndError=0) and AxIndErr to
      plot the various curves.  Finally got the ClError and AxIndError vs AlfaD
      (or AInd or AIndComputed) to cross at the same x-axis values!  I did this
      by solving for AxIndErr=0, solving for Dct, then solving Dct for Sw, then
      solving Sw for CnLoc, then solving CnLoc for ClLoc.  See 6-22-09 Notes.

3.02.00k05-dcm, 22-Jun-2009
 Removed screen output of convergence variables.

3.02.00m02-dcm, 7-Jul-2009
 Changed ClBem and DCTzero calculations.
 Added loop to calculate DctBem and new ClBem.
 The plots match what is in the text books now.

3.02.00m03-dcm, 7-Jul-2009
 Working with the singularity.  Ouput SinAF.

3.02.00n01-dcm, 7-Jul-2009
 New formula to calculate Dct.  Removes (1/(sin(phi)*sin(phi))) term.
      Let's see if it works!  New formula labeled as DctNew to compare with old.

3.02.00p01-dcm, 8-Jul-2009
 New TanInd formula.  Uses equation which assumes induced velocity perpendicular
      to relative velocity for areas near AInd=1.0 to avoid the singularity.
      This equation is smooth and continuous through AInd=1.0.  It gives similar
      results to the old equation except at low or high AInd.  Is Vi not
      perpendicular to W here?

3.02.00p02-dcm, 9-Jul-2009
 Applied TanInd equation from actuator disc theory.  Equation 5.10 in Wilson's
      1984 paper "Performance Analysis of HAWT" Oregon State Univ.  This
      equation is not a function of CtLoc.  The strength of the singularity at
      AInd=1.0 is due to CtLoc.

3.02.00p03-dcm, 9-Jul-2009
 Try new form of Tangential Induction Factor. (Form 3)  Removes Cd from
      calculation of Tangential Induction Factor which removes the singularity!

3.02.00q01-dcm, 16-Jul-2009
 New TanInd calculation.

3.02.00q04-dcm, 20-Jul-2009
 TanInd calculation using Amin and Amax with Cosine transtion from
      AminSing*(1.0-DelSing) to Amin and Amax to AmaxSing*(1.0+DelSing)

3.02.00q05-dcm, 20-Jul-2009
 Changed to new Dct calculation for 0.97 < AInd < 1.03 due to singularity still
      appearing during some extreme calculations.

3.02.00q06-dcm, 20-Jul-2009
 Last version with first version of singularity correction.

3.02.01a00-dcm, 27-Jul-2009
 Finished version of singularity correction.

3.03.01a00-dcm, 27-Jul-2009
 Changed IndType in input file from integer type to logical, since Choice 2
      (PROPX routine) has been removed.  Also, added TISingularity option to
      input file.

3.03.01a01-dcm, 27-Jul-2009
 Newton Raphson Algorithm

3.03.01a03-dcm, 28-Jul-2009
 Newton Raphson Algorithm Working

3.03.01a06-dcm, 29-Jul-2009
 Changed .dbg file output to only list non-converged cases.

3.03.01b01-dcm, 3-Aug-2009
 Added Find ZC (root bounding) algorithm.

!Temp End 3.03.01b03-dcm, 4-Aug-2009
 Moved TanInd calculation in subroutine GetAFTi to a call of Subroutine
      TanIndErr

3.03.01c01-dcm, 6-Aug-2009
 Added Binary routine to convergence methods, as well as a check for "jumping"
      between two points.

3.03.01c02-dcm
 Robust Binary search added.  Converges for all cases in UAE) test case (22
      wouldn't converge with 2004 version.  But, this version takes 104.19
      seconds vs 21.953 with 2004 version.

3.03.01d01-dcm, 7-Aug-2009
 Try NewtonRaphson, then Binary to speed up convergence.

v3.03.02a-mlb, 3-Dec-2009  M. Buhl
 Fixed the units for BED files when generating tab-delimited data
      [WTP_Subs.f90\RotAnal()].
 I modified WTP_Subs.f90\RotAnal() so that when generating tab-delimited BED
      files, the rotor-speed, pitch, and wind-speed (or TSR) values in the
      header were in separate cells so that people could compute using the
      values in Excel.
 I replaced all the calls to Read?Var to ReadVar because of the overloading now
      available in the library.

v3.03.02a-mlb, 4-Dec-2009  M. Buhl
 I added UnfPower to input file.  I added UnUn to WTP_Mods.f90\Parameters().  I
      added logic to output the power in unformatted form in
      WTP_Subs.f90\ParmAnal().

v3.03.02a-mlb, 10-Dec-2009  M. Buhl
 I added OutMaxCp to the input file.  It is a option to output the conditions
      that resulted in the maximum Cp.  I added the logic to compute it and
      include it in the output file.

v3.04.00a-mlb, 01-May-2010  M. Buhl
 I added the ability to handle airfoil files that include Cp,min.

v3.04.00b-mlb, 05-Nov-2010  M. Buhl
 I relinked it with the latest NWTC Library (v1.03.00) and changed the call to
      OpenCon to NWTC_Init.

v3.04.00c-mlb, 13-Jan-2011  M. Buhl
 Eliot Peterson of Clipper discovered that there was little shear effect being
      seen.  I found that, when dimensional input is used, the nondimensional
      span location was divided by the dimensional hub height instead of the
      nondimensional hub height.  I changed GetData() and RotAnal() to fix it.
 I relinked it with the latest NWTC Library (v1.03.01).

v3.04.00d-mlb, 21-Jun-2011  M. Buhl
 I fixed some coding issues related to compiling with gcc v4.5.3: All calls to
      Abort() were changed to ProgAbort().  This was not documented within the
      code because it affected so much!  I had to change the return variable for
      the recursive FindZC() function to "ZC" to avoid a compiler error.
 I fixed the problem with finding the conditions for maximum Cp.


v3.04.01a-dcm, 21-Jun-2010  D. Maniaci
v3.04.01b-dcm, 22-Jun-2010
 Adding Cavitation model and new inputs for CAV and DAWT models.

v3.04.01c-dcm, 24-Jun-2010
 Defined HubRadND and HubHtND for non-dimensional input; in GetData in WTP_Subs.
 Set "ShearLoc = ( 1.0 + HtLocND/HubHtND )**ShearExp" instead of
      "ShearLoc = ( 1.0 + HtLocND/HubHt )**ShearExp"
 Added DAWT correction factors:CavBeta*CavLambda.
      "VWndGnd  = VelHH*ShearLoc*CavBeta*CavLambda"
 UnDA = 9 ! Unit for diffuser augmented turbine analysis
 Added DAWT and Cavitation logical inputs to GetData.
 Added Cavitation Real inputs to GetData.
 Added above 6 variables to Module PropGen.
 Changed 'HubHt' to 'HubHtND' in GetData error check:
      "IF ( HubHtND*CosCone*CosTilt .LT. 1.0 )  CALL Abort ( ' The hub is so
      low, the blade will hit the ground.' )"
 Added checks for all new inputs.  No check for using a Diffuser on a wind
      turbine (yet).

v3.04.01d-dcm, 25-Jun-2010
 Changed FindZC from a function to a subroutine, had the returned value passed
      as an arguement.  Fixed error which showed up after applying more
      stringent debugger options.

v3.04.01g-dcm, 30-Jun-2010
 Changed out-nines (for non converged and no-solution cases) so that all
      parameters output 99's, not just Cp and Power.
 Changed out-sevens for same thing as previous line for cavitation.

v3.04.01h-dcm, 2-Jul-2010 D. Maniaci
 Changed GetCoefs() call to use logicals UseCm and UseCpmin to decide whether to
      read and return values for Cm and Cpmin.

v3.04.01i-dcs, 14-Jul-2011 Danny Sale
 I merged the changes made in (v3.04.00d-mlb, 21-Jun-2011  M. Buhl) into
   	(v3.04.01h-dcm, 2-Jul-2010 D. Maniaci), and then I continued to work off
      of Dave's v3.04.01h-dcm version
 Changed the call to OpenCon to NWTC_Init in WT_Perf.f90
 Added these additional variables to the .bed output file: local element radius,
      incidence angle (pitch+local twist), pitching moment coefficient, minimum
      pressure coefficient, cavitation number, cavitation flag (outputs true if
      cavitation is detected, false otherwise).
 Moved the initiation of CmLoc (the local pitching moment coef) to the WTP_Data
      modul.
 Renamed the variable "CavitationCheck" to "CavElement"
 Added a message to the .oup output file that indicates if cavitation occured or
      not.  If cavitation occured for any of the cases of a parametric analysis,
      there will be a message at the header of the file that says "Cavitation:
      Yes", and if cavitation did not occur then it will say "Cavitation: No".
      For a combined case analysis, there will be a seperate column that outputs
      "T" or "F" so we can distinguish which case(s) cavitation occured on.
      "T" means cavitaion occured, "F" otherwise
 Changed the way the cavitation safety factor is applied in the check for
      cavitation.  Now the cavitation safety factor is multiplied to the vapor
      pressure, I think this is more intuitive.
 If the user wants to run a cavitation analysis, I added a check to make sure
      that the user has input minimum pressure coefficients.
 For parametric anlysis, the header of the .oup file was being repeated at the
      top of each data table in the .oup output file.  I moved the write
      statements for the .oup header outside of a DO loop, so that now it only
      gets written once at the top of the file.
 I modified the alignment of the sectors for sheared flow analysis. Previously,
      it was never possible to analyze the "12-o-clock" position due to the way
      the sectors were defined (the azimuth angle could never reach 0 degrees),
      but now the first azimuth angle starts at 0 degrees.  The motivation for
      this modification was to check for cavitation at the "12-o-clock"
      position, and also this should be more accurate for sheared inflow in
      general.
 Updated the formula for HtLocND (the non-dimensional height of the blade
      segment above the hub elevation).  This new formula was suggested by
      Gunjit, who suggested it to Marshall, who suggested it to me, and it is
      suggested that this new formula is slightly more acurate ;)
 I updated the formula for atmospheric pressure (PressAtm) and defined a new
      variable for the gravitational acceleration (Gravity) in WTP_Data.  The
      new variable Gravity works for both sets of units (English and Metric).
 I added some additional logic in the IF statement which checks for cavitation.
      Namely, the cavitation flags were not being reset at the correct times,	so
      I fixed this problem and added some dummy values for when cavitation
      analysis is not requested by the user.
 Commented out the statements which output "Sevens" for when cavitation is
      detected.  Instead, I added "True" or "False" flags to be output when
      cavitation is (or is not) detected, but the originally calculated values
      for power, power coefficient, etc... are still output as normal. But look
      at the cavitation flags in the output files so you know whether or not you
      can trust the predictions!
 I think the variable SpdUns was being truncated, I made a small change to the
      assignment of SpdUns and the compiler stopped complaining now
 I updated all of the sample input files in the CertTest folder to be compatible
      with this current alpha version of WT_Perf, and I added a new test case
      named "TEST07_MHK_RefModel" which represents a hydrokinetic turbine and
      utilizes the new cavitation analysis. I updated ArcFiles.txt in the
      developer's archive to include the files from this new test case.
 I compiled this alpha version of WT_Perf with gfortran v4.7 dated 2011-04-19,
      and it works!  I also tested to see if the g95 compiler works, and it does
      too!  I linked this version with the alpha NWTC Subroutine Library
      (v1.04.00c).
 I updated the user's guide to reflect this current alpha version of WT_Perf.


v3.04.01j-dcs, 20-Jul-2011 Danny Sale
 This is a fork of v3.04.01i-dcs that includes a cascade model to account for
      blades passing through the wakes of previous turbines.  It will likely be
      merged back into the code base later after a release.


v3.05.00a-adp, 09-Nov-2012 Andy Platt
 This is a reformatted version of v3.04.01i-dcs. Indentation and tabbing were
      changed for readability.  Since the input file changed from v3.04.01h
      to v3.04.01i, we incriment the second number set in this version.
 I included a minor change in the BinSearch subroutine of InductBEM so that it
      now runs for NSplit times before exiting (used to be hard coded at 10
      iterations).
 There was a convergence issue related to the 2-D Newt Raphson routine where it
      is calculation both AxInd and TanInd at the same time, it changes the
      value of TanInd. If it does not find a solution, it exits leaving TanInd
      with the last value it tried (Which could be of order of hundreds -- which
      is non-physical). The value of TanInd is reset to some initial guess after
      the Newt-Raphson iterations exit without finding a solution.
 Removed the DAWT feature since this was never fully developed (it was merely a
      multiplicative on the area of 1.3 -- which effectively multiplied the
      input velocity). The flag is still read from the input, but now the
      program will present an error message and abort. See the programming and
      users guides for more inforation.
 Removed the failed cases counters within the loops of the RotAnal subroutine.
      They were causing inacurate reporting of the number of actual cases run
      that did not converge (i.e. of 1 case analyzed, 9 failed...).
 In the subroutine InductBEM, I changed the lines that read "Solution = .TRUE."
      to read "ZFound   = .TRUE.".  The reason is that ZFound is passed out to
      the RotAnal subroutine into the variable Solution.  Previously it was
      possible to set Solution to true, but then pass out a false ZFound that
      overwrote it.
 Modified the BED output for non-converging cases. Now it sets the cavitation
      flag as false (used to be set as true). This avoids having a blade element
      data (bed) file showing cavitation when the output (oup) file does not.
 In WTP_Subs.f90, replaced the Flt2LStr function with the more generic Num2LStr
      function. The latter command is an interface within the NWTC Library. It
      points to the Flt2LStr function as required.
 Tweaked the limits imposed on TanIndErr for when it runs and what it can return
      as viable answers.  Tweaked the limits on the NewtRaphson for what values
      it can return.  This improved the stability of the code and sped it up
      slightly.
 Added in a flag to the oup file (for combined cases only) and a flag to the bed
      file to indicate if the case completely converged.  This makes it easier
      to see if the results worked properly.  This required some restructuring
      of how the RotAnal routine works.
 Replaced the "OutNines" option with ConvFlag to specify how to handle non-
      convergent cases.  This takes an option of 0 (output the bad result),
      1 (the old outnines behaviour), 2 (NaN).
 Changed the exiting of WT_Perf for analysis of a single non-convergent case.
      Now reports normal exit rather than aborting.
 Changed the way NSplit is handled as suggested by Danny Sale. This is now read
      in from the input file and treated as a parameter. The lower limit of 1 is
      enforced, but no upper limit is in place.
 Modified the CertTests for linux compatability.
 Added a makefile for use on linux systems.


======
To Do:
 Add user-defines output format
 Add DAWT analysis -- Read Beta and Lamdbda terms from a new input file. (Like
      reading airfoil data, in GetAF).
 Add convergence vs solution check (so the user knows if there is a solution,
      but no convergence.)
 Add Converge to the BinSearch argument list instead of treating it as global.
 Add AeroDyn initial axial induction factor calculation (Equation 20 in AeroDyn
      theory manual).  May or may not help induciton iteration.
 Add Cq and Ct output.
 Add input to specify output format.
 Make compatible with AeroDyn 13.
 Try to make induction of a node close in value to adjacent nodes.
 Update the manual to describe the process of creating analysis nodes.
 Option to eliminate headers for easy import into MatLab.
 Consider using previous, converged values (for a given case) as initial guesses
      and maybe forget the exhaustive search unless the N-R fails to converge.
 Include the algorithm to interpolate lift & drag coefficients from the new
      AirfoilPrep algorithm (v2.3) added by Danny Sale