Friday, July 14, 2017

Conduits in #SWMM5

Conduits
Conduits are pipes or channels that move water from one node to another in the conveyance system. Their cross-sectional shapes can be selected from a variety of standard open and closed geometries as listed in the following table. Irregular natural cross-section shapes and Dummy links are also supported.
SWMM5 Lnk Shape

Conduits are pipes or channels that move water from one node to another in the conveyance system. Their cross-sectional shapes can be selected from a variety of standard open and closed geometries as listed in Table 3-1.

Most open channels can be represented with a rectangular, trapezoidal, or user-defined irregular cross-section shape. For the latter, a Transect object is used to define how depth varies with distance across the cross-section (see Section 3.3.5 below). Most new drainage and sewer pipes are circular while culverts typically have elliptical or arch shapes. Elliptical and Arch pipes come in standard sizes that are listed in at the bottom of this page. The Filled Circular shape allows the bottom of a circular pipe to be filled with sediment and thus limit its flow capacity. The Custom Closed Shape allows any closed geometrical shape that is symmetrical about the center line to be defined by supplying a Shape Curve for the cross section (see Section3.3.11 below).

SWMM uses the Manning equation to express the relationship between flow rate (Q), crosssectional area (A), hydraulic radius (R), and slope (S) in all conduits. For standard U.S. units,

where n is the Manning roughness coefficient. The slope S is interpreted as either the conduit slope or the friction slope (i.e., head loss per unit length), depending on the flow routing method used. 

For pipes with Circular Force Main cross-sections either the Hazen-Williams or Darcy-Weisbach formula is used in place of the Manning equation for fully pressurized flow. For U.S. units the Hazen-Williams formula is:
where C is the Hazen-Williams C-factor which varies inversely with surface roughness and is supplied as one of the cross-section’s parameters. The Darcy-Weisbach formula is:

where g is the acceleration of gravity and f is the Darcy-Weisbach friction factor. For turbulent flow, the latter is determined from the height of the roughness elements on the walls of the pipe (supplied as an input parameter) and the flow’s Reynolds Number using the Colebrook-White equation. The choice of which equation to use is a user-supplied option.

A conduit does not have to be assigned a Force Main shape for it to pressurize. Any of the closed cross-section shapes can potentially pressurize and thus function as force mains that use the Manning equation to compute friction losses. 

A constant rate of exfiltration of water along the length of the conduit can be modeled by supplying a Seepage Rate value (in/hr or mm/hr). This only accounts for seepage losses, not infiltration of rainfall dependent groundwater. The latter can be modeled using SWMM’s RDII feature (see Section 3.3.6).

The principal input parameters for conduits are:
  • names of the inlet and outlet nodes
  • offset heights of the conduit above the inlet and outlet node inverts
  • conduit length
  • Manning’s roughness
  • cross-sectional geometry
  • entrance/exit losses
  • presence of a flap gate to prevent reverse flow.
A conduit can also be designated to act as a culvert (see Figure 3-2) if a Culvert Inlet Geometry code number is assigned to it. These code numbers are listed in Appendix A.10. Culvert conduits are checked continuously during dynamic wave flow routing to see if they operate under Inlet Control as defined in the Federal Highway Administration’s publication Hydraulic Design of Highway Culverts Third Edition (Publication No. FHWA-HIF-12-026, April 2012). Under inlet control a culvert obeys a particular flow versus inlet depth rating curve whose shape depends on the culvert’s shape, size, slope, and inlet geometry. 

Flow Regulators
Flow Regulators are structures or devices used to control and divert flows within a conveyance system. They are typically used to:
  • control releases from storage facilities
  • prevent unacceptable surcharging
  • divert flow to treatment facilities and interceptors
 InfoSWMM H2OMap SWMM InfoSWMM SA  can model the following types of flow regulators:
  • Orifices
  • Weirs
  • Outlets
The following Tables are copied from the EPA Manual on SWMM (Hydraulics) II




Wednesday, July 12, 2017

SWMM5_NR_ITERATIVE Fortran Routine from 2004

      SUBROUTINE SWMM5_NR_ITERATIVE
C EXTRAN BLOCK test for SWMM 5 beta solution - 12/12/2002
      INCLUDE 'TAPES.INC'
      INCLUDE 'STIMER.INC'
      INCLUDE 'BD.INC'
      INCLUDE 'BND.INC'
      INCLUDE 'HYFLOW.INC'
      INCLUDE 'CONTR.INC'
      INCLUDE 'JUNC.INC'
      INCLUDE 'PIPE.INC'
      INCLUDE 'TIDE.INC'
      INCLUDE 'OUT.INC'
      INCLUDE 'ORF.INC'
      INCLUDE 'WEIR.INC'
      INCLUDE 'FLODAT.INC'
      DOUBLE PRECISION AKON,QNEW,DELQ1,DELQ2,DELQ3,DELQ4,DELQ5
DIMENSION        AS1(NEE)
DOUBLE PRECISION df,f,n_omega
integer          good_nodes
C=======================================================================
C     STORE OLD TIME STEP FLOW VALUES
C=======================================================================
DO  N        = 1,NTL
      QO(N)        = Q(N)
      AT(N)        = A(N)
      VT(N)        = V(N)
enddo
C=======================================================================
C     INITIALIZE CONTINUITY PARAMETERS
C=======================================================================
DO J            = 1,NJ
if(othercom(63).eq.1) then
                     Y(J)      = Y(J) + 0.5 * ( YEX2(J) - 
     +                                  YEX1(J) + YEX1(J) - YO(J))
                     IF(Y(J).LT.FUDGE)      Y(J)=FUDGE
                     IF(Y(J).GT.SURELEV(J)) Y(J)=SURELEV(J)-Z(J)
                     yex2(j)   = yex1(j)
                     yex1(j)   = YO(j)
                     endif
cred  beginning time step value of the node area
asold(j)    = as(j)
      YO(J)       = Y(J) 
      GoodNode(j) = .FALSE.
enddo
good_nodes  = 0
      omega       = input_omega
      n_omega     = node_omega
loop_count  = 0
C=======================================================================
C     HALF-STEP AREA, RADIUS : VELOCITY
C     FULL-STEP FLOW
C=======================================================================
big_loop:  DO while(good_nodes.lt.nj.and.loop_count.le.itmax)
loop_count = loop_count + 1
      if(loop_count.ge.itmax-5) then
        omega   = 0.50 * omega
        n_omega = 0.50 * n_omega 
        endif

DO J           = 1,NJ
      AS(J)          = AMEN
      AS1(J)         = 0.0
      SUMQ(J)        = QIN(J)
      SUMQS(J)       = QIN(J)
      SUMAL(J)       = 0.0
enddo
CIM   FIRST COMPUTE GATED ORIFICE PARAMETERS
      CALL OGATES(DELT,Y,V)
c
      flow_loop: DO    N      = 1,NTC
      NL            = NJUNC(N,1)
      NH            = NJUNC(N,2)
C=======================================================================
      H(N,1)   = AMAX1(Y(NL) + Z(NL),ZU(N))
      H(N,2)   = AMAX1(Y(NH) + Z(NH),ZD(N))
      CALL nhead(N,NL,NH,H(N,1),H(N,2),Q(N),A(N),V(N),HRAD,
     +           ANH,ANL,RNL,RNH,YNL,YNH,width,IDOIT,LINK(N),AS1)
cred  bypass loop for nodes already converged
      bypass_loop: if(loop_count.gt.2.and.
     +  goodnode(nl).eq..TRUE..and.goodnode(nh).eq..TRUE.) then
bypass = bypass + 1.0
else
IF(HRAD.GT.HMAX(N)) HMAX(N) = HRAD
      IF(A(N).GT.AMAX(N)) AMAX(N) = A(N)
cred  save information for the culvert classification
      HRLAST(N)  = HRAD 
vup(n)     = anl
vdn(n)     = anh
if(loop_count.eq.1) then
                   aup(n)   = anl
                   rup(n)   = rnl
                   rmd(n)   = hrad
                   endif
      positive_flow: IF(IDOIT.gt.0) THEN
c
c       Q/ANH = velocity at downstream end used for exit loss
c       Q/ANL = velocity at upstream end used for entrance loss
c       The loss = 1/2*K*V^2/g is factored into momentum
c       equation similarly to the friction slope
c       The loss momentum term = g*A*[1/2*K*V^2/g]
c                              = g*A*[1/2*K*Q/A*Q/A/g]
C                               =g  *[1/2*K*|Q*A|  /g] * Q
C                               =    [1/2*K*|Q*A|    ] * Q
c       DELQ5 is sum of losses
c
cred  calculate the froude number for every conduit
c
      DELH      = H(N,1) - H(N,2)
      DELZP     = ZU(N)  - ZD(N)
      DIFF_mid  = 0.5 * (H(N,1) - ZU(N) + H(N,2) - ZD(N))
      IF(DIFF_mid.GT.0.0) THEN
           FROUDE_MID = ABS(Q(N))/A(N)/SQRT(GRVT*(DIFF_mid))
                  ELSE
           FROUDE_MID = 0.0
           ENDIF
      bfactor = 0.0
if(FROUDE_MID.ge.1.0) THEN
           bfactor = 0.0
           elseif(froude_mid.gt.0.9) then
bfactor = 1.0 - (10.0*FROUDE_MID-9.0)
           endif
cred  test for zero sloped conduits
      if(delzp.eq.0.0)     then
                    del_ratio   = delh/0.001
                    else
                    del_ratio   = delh/delzp
                    endif
cred  swmm 4 definition for normal flow
      if(del_ratio.le.1.0) then 
                    delfactor = 0.0
cred                       swmm 5 transition definition
                    else if(del_ratio.gt.1.10) then
delfactor = 1.0
                    else
                    delfactor = 10.0 * (del_ratio-1.0)
                      endif
      DELQ5          = 0.0
      IF(ANH.NE.0.0)   DELQ5 = DELQ5 + 0.5 * ABS(Q(N)/ANH)*ENTK(N)
      IF(ANL.NE.0.0)   DELQ5 = DELQ5 + 0.5 * ABS(Q(N)/ANL)*EXITK(N)
      IF(A(N).NE.0.0)  DELQ5 = DELQ5 + 0.5 * ABS(Q(N)/A(N))*OTHERK(N)
      DELQ5 =  DELQ5 * DELT/LEN(N)
c
      DELQ4  = DELT*V(N)*(ANH-ANL)/A(N)/LEN(N)                                          
cred  the mean area travels from the midpoint to the upstream area
cred  as the value of delh/delzp changes
      area_mean =  wt*anl + wd*aup(n)  + ( wt*a(n) + wd*at(n) - 
     +             wt*anl - wd*aup(n)) *   delfactor

      DELQ2  = DELT*GRVT*area_mean*((H(N,2) - H(N,1)))/LEN(N)
      DELQ3  = 2.0*(A(N)-AT(N))/A(N)
cred  the mean hydraulic radius travels from the midpoint to the 
cred  upstream hydraulic radius as the value of delh/delzp changes
      hrad_mean =  wt*rnl + wd*rup(n)  + ( wt*hrad + wd*rmd(n) - 
     +             wt*rnl - wd*rup(n)) *   delfactor
      DELQ1  = DELT*(ROUGH(N)/hrad_mean**1.33333)*ABS(V(N))

QNEW  = QO(N) - delq2
      AKON  = DELQ1 + DELQ5 - delq4*bfactor - DELQ3*bfactor
cred  Newton-Raphson iteration
      F         = Q(N) * ( 1.0 + akon) - qnew
DF        = 1.0 + akon
Q(N)      = Q(N) - omega*f/df
      DQDH      = 1.0/(1.0+AKON)*GRVT*DELT*A(N)/LEN2(N)
dqdh_old(n) = dqdh
C=======================================================================
C     DO NOT ALLOW A FLOW REVERSAL IN ONE TIME STEP
C=======================================================================
      DIRQT = SIGN(1.0,QO(N))
      DIRQ  = SIGN(1.0,Q(N))
      IF(DIRQT/DIRQ.LT.0.0) Q(N) = 0.001*DIRQ

v(n)   = q(n)/A(N)
C=======================================================================
C     COMPUTE CONTINUITY PARAMETERS
C=======================================================================
      SELECT CASE (INGATE(N))
      CASE (1)
      Q(N) = AMAX1(Q(N),0.0)
      CASE (2)
      Q(N) = AMIN1(Q(N),0.0)
      END SELECT

      IF (NKLASS(N).LE.21) THEN
                      Q(N) = AMAX1(STHETA(N),Q(N))
                           Q(N) = AMIN1(SPHI(N),Q(N))
                    ENDIF
endif positive_flow
      endif bypass_loop
      SUMQS(NL) = SUMQS(NL) - Q(N)*barrels(n)
      SUMAL(NL) = SUMAL(NL) + dqdh_old(n)*barrels(n)
      SUMQS(NH) = SUMQS(NH) + Q(N)*barrels(n)
      SUMAL(NH) = SUMAL(NH) + dqdh_old(n)*barrels(n)

      SUMQ(NL)   = SUMQ(NL)  - WT*Q(N)*barrels(n)  - WD*QO(N)*barrels(n)   
      SUMQ(NH)   = SUMQ(NH)  + WT*Q(N)*barrels(n)  + WD*QO(N)*barrels(n) 
      ENDDO  flow_loop
C=======================================================================
C     SET FULL STEP OUTFLOWS AND INTERNAL TRANSFERS
C=======================================================================
      CALL BOUND(Y,Y,Q,TIME,DELT)
C=======================================================================
      N1       = NTC+1
      DO 370 N = N1,NTL
      NL       = NJUNC(N,1)
      NH       = NJUNC(N,2)
      IF(ABS(Q(N)).LT.1E-10) Q(N) = 0.0
C=======================================================================
      SUMQ(NL)  = SUMQ(NL)  - WT*Q(N)*barrels(n)  - WD*QO(N)*barrels(n) 
      SUMQS(NL) = SUMQS(NL) - Q(N)*barrels(n)    
      IF(NH.NE.0) THEN
         SUMQ(NH)  = SUMQ(NH) + WT*Q(N)*barrels(n) + WD*QO(N)*barrels(n) 
         SUMQS(NH) = SUMQS(NH) + Q(N)*barrels(n)    
         ENDIF
  370 CONTINUE
C=======================================================================
C     CALCULATE THE FULL-STEP HEAD
C=======================================================================
      DO J = 1, NJ
         AS(J)    = AS(J)  + AS1(J) 
      ENDDO
      DO  J   = 1,NJ
      IF(JSKIP(J).le.0) then
C=======================================================================
cred    time weighted area average
      if(othercom(62).eq.1) then
                average_area = WT*as(j) + WD*asold(j)
                else
                average_area = as(j)
                endif
cred    Newton-Raphson iteration
        yt(j)     = y(j)
 if(y(j).le.ycrown(j)) then
 F         =  Y(J) * average_area - YO(J) * average_area   
     +                    - sumq(j)*DELT
   DF        =  average_area 
        ASFULL(J) = AS(J)
 else
   DF        =  sumal(j) 
        IF(Y(J).LT.1.25*YCROWN(J))  DF = DF + 
     +   (ASFULL(J)/DELT-SUMAL(J))*EXP(-15.*(Y(J)-YCROWN(J))/YCROWN(J))
cred       correction from SWMM 5 QA testing - 11/12/2004
cred       if a large delt or if there is a large value of sumal(j)
cred       - usually from a very large conduit connected to the current node - 
cred       the expression  ASFULL(J)/DELT-SUMAL(J) may be negative 
cred       invalidating the whole concept of a "transition" slot
           if(Y(J).LT.1.25*YCROWN(J).le.ASFULL(J)/DELT.le.SUMAL(J)) 
     +     denom = sumal(j)
        CORR     = 1.00
        IF(NCHAN(J,2).EQ.0) CORR  = 0.60
 F       =  Y(J) * DF - YO(J) * DF - CORR * sumqs(j)
cred    WRITE(*,*) ajun(j),df,f,y(j),ycrown(j),sumal(j)
 endif
     Y(J)    =  Y(J) - n_omega*F/DF
        IF(Y(J).LT.0.0) Y(J) = 0.0
        IF((Y(J)+Z(J)).GT.SURELEV(J)) Y(J) = SURELEV(J)-Z(J)
 endif
       enddo

cred  converge until all nodes meet the convergence criteria
good_nodes = 0 
error_node = 0
i_was_bad  = 1 
DO j       = 1,nj
if(jskip(j).le.0) then
                  if(abs(y(j)-yt(j)).le.node_toler) then
                                     good_nodes = good_nodes + 1
                                     GoodNode(j) = .TRUE.
                              endif
                   if(abs(y(j)-yt(j)).gt.error_node) then
       error_node = abs(y(j)-yt(j))
                          i_was_bad  = j
                   endif
                 else
                  good_nodes  = good_nodes + 1
           GoodNode(j) = .TRUE.
endif
enddo 
c     write(*,669) loop_count,nj-good_nodes,error_node,
c    +               delt,ajun(i_was_bad),y(i_was_bad),  
c    +               sumq(i_was_bad),omega
669   format(2I6,F10.4,F8.2,1x,a12,3F9.3)
enddo big_loop

cred  culvert information - 8/6/2002
cred  culvert information for the culvert comparison file - 8/6/2002
cred  culvert information - 8/6/2002
      if(othercom(90).eq.1) then
       do n = 1,nc
       if(nklass(n).eq.1.or.nklass(n).eq.21) then
       if(abs(q(n)).gt.culvert_loss(n,5).and.abs(v(n)).lt.20.0)then
                    HW          = H(N,1)   - ZU(N)
                    TW          = H(N,2)   - ZD(N)
                    vup(n)      = v(n)
                    vdn(n)      = v(n)
                    head_loss   = 0.5*ENTK(N)*vup(N)*vup(N)/GRVT  +
     +                            0.5*EXITK(N)*vdn(N)*vdn(N)/GRVT +
     +                            0.5*OTHERK(N)*v(N)*v(N)/GRVT 
                    sfloss  = ROUGH(N)/GRVT * 
     +                        abs(v(N))*ABS(v(N))/hrlast(n)**1.33333
cred                sfloss  = ROUGH(N)/GRVT * 
cred +                   Q(N)*ABS(Q(N))/(A(N)**2*HRLAST(N)**1.33333)
             culvert_loss(n,1) = len(n)*sfloss
             culvert_loss(n,2) = head_loss
             culvert_loss(n,3) = h(n,1)
             culvert_loss(n,4) = h(n,2)
             culvert_loss(n,5) = q(n)  
   endif 
             CALL DEPTHX(N,NKLASS(N),q(n),YC,YNORM)
cred                type 1     
                    IF(HW.LT.1.5*DEEP(N).AND.YC.LT.YNORM.and.
     +                         TW.LE.YC) then
                               culverted(N,1) = culverted(N,1) + DELT
                        culverted(n,8) = 1
                        endif
cred                type 2
                    IF(HW.LT.1.5*DEEP(N).AND.YC.LT.YNORM.AND.
     +                         TW.GT.YC.AND.TW.LE.DEEP(N)) then
                               culverted(N,2) = culverted(N,2) + DELT
                        culverted(n,8) = 2
                        endif
cred                 type 3
                     IF(YC.GE.YNORM.AND.TW.LT.ZU(N)-ZD(N)) then
                               culverted(N,3) = culverted(N,3) + DELT
                        culverted(n,8) = 3
                        endif
cred                 type 4
                     IF(YC.GE.YNORM.AND.TW.GE.ZU(N)-ZD(N)+YC.
     +                         AND.TW.LT.ZU(N)-ZD(N)+DEEP(N)) then
                               culverted(N,4) = culverted(N,4) + DELT
                        culverted(n,8) = 4
                        endif
cred                 type 5
                     IF(YC.LT.YNORM.AND.TW.GE.DEEP(N)) then
                               culverted(N,5) = culverted(N,5) + DELT
                        culverted(n,8) = 5
                        endif
cred                 type 6
                     IF(YC.GE.YNORM.AND.TW.GE.ZU(N)-ZD(N)+
     +                         DEEP(N)) then
                               culverted(N,6) = culverted(N,6) + DELT
                        culverted(n,8) = 6
                        endif
cred                 type 7
                     IF(HW.GE.1.5*DEEP(N).AND.TW.LT.DEEP(N)) then
                               culverted(N,7) = culverted(N,7) + DELT
                        culverted(n,8) = 7
                        endif
                      endif
                      enddo 

      endif
      RETURN
      END

Monday, July 10, 2017

The relationship between the current link mid point velocity and the time step in a link of #SWMM5

SWMM5 uses the Courant–Friedrichs–Lewy (CFL) condition to compute the variable time step used during each time step of a simulation.   In general, the shortest link length with the highest velocity and depth will dominate the time step computations.   The CFL time step is the link length over the current velocity plus the current wave celerity.  If you plot the velocity and computed time step over time the time step will decrease as the velocity increases (Figure 1).  Figure 1 also shows why a hot start file is important in SWMM5.  If you start out with a dry network the SWMM5 engine will not be able to have a good estimate of the needed CFL time step.  The depth and velocity will be zero.

Figure 1 - The relationship between the current link mid point velocity and the time step in a link of SWMM5.

Sunday, July 9, 2017

#SWMM3 and #SWMM4 Acknowledgments, includes mention of #PCSWMM

A note from the past, the SWMM3 and SWMM4 Acknowledgments from their respective manuals for the sake of posterity and in thanks for all of their past efforts.

SWMM4 Acknowledgments

Maintenance and updating of the EPA SWMM has been continuous since its
inception in 1969-70.

Over the many intervening years many individuals have contributed to its improvement, notably EPA colleagues Mr. Richard Field, Mr. Harry Torno, Mr. Chiu-Yuan Fan, Mr. Doug Ammon and Mr. Tom Barnwell. Mr. Torno and Mr. Barnwell have also managed the Storm and Water Quality Model Users Group (formerly the SWMM Users Group), a source of invaluable feedback from model users, including a large contingent from Canada and abroad.

Too many individuals have contributed to specific improvements to list here. This users manual, however, is based upon earlier versions to which the following persons contributed significant authorship while at the University of Florida: Mr. Brett A. Cunningham, Mr. Victor Gagliardo, Dr.Stephen J. Nix, Mr. Donald J. Polmann and Mr. W. Alan Peltz. In particular, Mssrs. Cunningham and Gagliardo developed the new subsurface routing routine in the Runoff Block. Dr. James P.Heaney has served staunchly as colleague, critic, and pioneer of new ideas. Omission of these names from the current list of authors does not diminish our gratitude for current and past efforts in developing the model.

The Extran Block is one of the most valuable and widely-used components of SWMM. Dr. Larry A. Roesner and Mr. John A. Aldrich of Camp, Dresser and McKee, Inc., one of the three original SWMM developers, have given generously of their time to enhance Extran and to provide useful suggestions for improvements of Extran and the rest of the SWMM model. The Fortran-77 code for Version 4 of SWMM is based on a microcomputer version prepared by Mr. Richard M. Baker and Mr. Karl J. Brazauskas of Metcalf and Eddy, Inc., another one of the original three developers. Much of the user’s manual text for Version 4 has been adapted from a computerized edition of the Version 3 manuals prepared by Dr. William James of Wayne State University. We are grateful to Dr. James and to Dr. Stephen Nix of Syracuse University for their helpful comments regarding Version 4. Assistance in printing of the manuals was provided by KBN Applied Sciences and Engineering, Inc. of Gainesville.

At the University of Florida, invaluable word-processing and SWMM dissemination duties have been performed faithfully by Ms. Doris Smithson. Main-frame computations were performed at the Northeast Regional Data Center on the University of Florida campus, Gainesville.

SWMM3 Acknowledgements

Maintenance and updating of the EPA SWMM has been continuous since its
inception in 1969-70. Over the several intervening years, many
individuals have contributed to its improvement, most notably EPA
col­leagues Richard Field, Harry Torno, Chi-Yuan Fan, Doug Ammon and
Tom Barnwell. Harry Torno and Tom Barnwell have also managed the SWMM
Users Group, through which many helpful suggestions for improvements
have come, including those from the large contingent of Canadian
users.

Regarding specific components of SWMM Version III, the Green-Ampt
infiltration routines were reviewed, programmed and tested by Dr.
Russell G. Mein, Department of Civil Engineering, Monash University,
Clayton, Victoria, Australia while on a sabbatical at the University
of Florida. He also provided valuable review and testing of other
model components. The earliest implementation of continuous simulation
in the Runoff and Storage/Treatment Blocks was done by George F.
Smith, now with the Office of Hydrology, National Weather Service,
Silver Spring, Maryland. Basic formulation of the snowmelt routines
was done following the work of Proctor and Redfern, Ltd. and James F.
MacLaren, Ltd., Toronto, who were under contract to the Ontario
Ministry of the Environment and the Canadian Environmental Protection
Service. Runoff Block surface quality changes were the subject of
masters research at the University of Florida by Douglas C. Ammon, now
with the EPA, Storm and Combined Sewer Branch, Cincinnati. Revision of
Transport Block scour/deposition routines is based on work with Dennis
Lai, Clinton-Bogert Associates, Fort Lee, New Jersey. Many lasting
improvements in SWMM programming were made by W. Alan Peltz, now with
General Electric, Atlanta.

Several others contributed to changes in the model. The card ID system
and the user-defined default values and ratios were suggested by the
Corps of Engineers, Hydrologic Engineering Center, Davis, California.
The programming basis has been aided by Dr. William James, McMaster
University, Hamilton, Ontario and exposure to his FASTSWMM programs.
Emphasis upon proper use and objectives of SWMM modeling has been
enhanced by conversations with the late Murray McPherson, Marblehead,
Massachusetts, Eugene Driscoll, Oakland, New Jersey, Dr. Dominic
DiToro, Manhattan College, New York City, John Mancini, Lincoln,
Nebraska, Dr. Paul Wisner, Ottawa University, Charles Howard,
Vancouver, B.C., and several others. OF is additionally grateful to
Reinhard Sprenger, Templeton Engineering, Winnipeg, for improvements
to the Extran Block, to Christian Eicher, Gore and Storrie, Ltd.,
Toronto, for several important corrections to the overall program, to
Robert Johnson, Lehigh University, Bethlehem, Pennsylvania, for
comments on the compatibility with CDC machines, to Tom Jewell, Union
College, Schenectady, New York, for analysis of surface washoff and
other comments, and to Tom Meinholz and Richard Race, formerly of
Envirex, Inc., Milwaukee, for suggestions on making the program more
suitable to prototype configurations.

The Extended Transport Block has been an invaluable addition to the
SWMM package. Developed by Water Resources Engineers (now a part of
Camp, Dresser and McKee), Extran may be the most widely used portion
of SWMM. Dr. Larry Roesner and the late Dr. Robert Shubinski of CDM,
Annan­dale, Virginia have given generously of their time in enhancing
Extran and in making other useful suggestions to SWMM modeling.

At the University of Florida, salutary programming and testing has
been conducted by J. Jay Santos, Efi Foufoula, Michael Kennedy, Kelly
Nead and Christina Neff. Typing has been performed by Linda Trawick,
Jeanette Heeb, Kim Karr and the College of Engineering Word Processing
Center. Figures were drafted by Terri Schubert, Micky Hartnett and
Anelia Crawford. Computations were performed at the Northeast Regional
Data Center on the University of Florida campus, Gainesville.

Wednesday, July 5, 2017

How does the CFL time Step Change in one SWMM 5 Link? #Inside_SWMM5

Introduction – the reason for these series of blogs are as an expanded view of the input, engine and output of #SWMM5 It is a companion to the EPA Documentation which I describe here:
I have noticed based on email questions and postings to the SWMM List Sever (a great resource hosted by CHI, Inc.) that many SWMM 5 users do not know about the really outstanding documentation on SWMM 5 posted on the EPA Website https://www.epa.gov/water-research/storm-water-management-model-swmm It consists of two now and in the near future three volumes on Hydrology, Water Quality, LID's and SuDS and Hydraulics. The documentation is fantastically complete with detailed background on the theory, process parameters and completely worked out examples for all of the processes in SWMM5. It is truly an outstanding aid to modelers and modellers worldwide. It would benefit you to read them (if you have not already downloaded the PDF files)

If you use variable time steps in SWMM5 then based on the CFL time step condition the program will adjust the time step.  The smallest link time step governs the program time step.  You can see the results for one link in Figure 1.  The time step increases as the velocity and depth increase during the simulation.
Variable Time Step (SWMM5 Help File)
Check the box if an internally computed variable time step should be used at each routing time period and select an adjustment (or safety) factor to apply to this time step. The variable time step is computed so as to satisfy the Courant condition within each conduit. A typical adjustment factor would be 75% to provide some margin of conservatism. The computed variable time step will not be less than the minimum variable step discussed below nor be greater than the fixed time step specified on the Time Steps page of the dialog.
        Figure 1.  The time step increases as the velocity and depth increase during the simulation.
                                                      

Tuesday, July 4, 2017

What is the minimum depth in a link of #SWMM5 ?

Introduction – the reason for these series of blogs are as an expanded view of the input, engine and output of #SWMM5 It is a companion to the EPA Documentation which I describe here:
I have noticed based on email questions and postings to the SWMM List Sever (a great resource hosted by CHI, Inc.) that many SWMM 5 users do not know about the really outstanding documentation on SWMM 5 posted on the EPA Website https://www.epa.gov/water-research/storm-water-management-model-swmm It consists of two now and in the near future three volumes on Hydrology, Water Quality, LID's and SuDS and Hydraulics. The documentation is fantastically complete with detailed background on the theory, process parameters and completely worked out examples for all of the processes in SWMM5. It is truly an outstanding aid to modelers and modellers worldwide. It would benefit you to read them (if you have not already downloaded the PDF files)
What it is the minimum depth in a link of SWMM5 ?   EPA SWMM5 uses internal units of feet or inches for rainfall.   The minimum depth is FUDGE in the Code and has the value of 0.001 feet or ft^2.  You can see the effect in the output file if you look at the depth, volume or area values.






AI Rivers of Wisdom about ICM SWMM

Here's the text "Rivers of Wisdom" formatted with one sentence per line: [Verse 1] 🌊 Beneath the ancient oak, where shadows p...