Description
A gas pipeline is to be designed to transport a fixed amount of gas from one point to two others. Both, the initial and final stages of the gas are known. We need to determine: Number of compressor stations Length of the pipeline segments Diameter of pipeline segments Suction and discharge pressures at each stations
Small Model of Type : MINLP
Category : GAMS Model library
Main file : gasnet.gms
$title Optimal Design of a Gas Transmission Network (GASNET,SEQ=223)
$onText
A gas pipeline is to be designed to transport a fixed amount of gas from
one point to two others. Both, the initial and final stages of the gas
are known. We need to determine:
Number of compressor stations
Length of the pipeline segments
Diameter of pipeline segments
Suction and discharge pressures at each stations
Edgar, T F, Himmelblau, D M, and Lasdon, L S, Optimization of
Chemical Processes. McGraw Hill, Boston, 2001.
Keywords: mixed integer nonlinear programming, pipeline designment, engineering
$offText
$eolCom //
Set
ii 'all pipe segments' / s0*s11 /
i(ii) 'pipe segments' / s1*s11 /
i1(i) 'segments one' / s1*s3 /
i2(i) 'segments two' / s4*s7 /
i3(i) 'segments three' / s8*s11 /
j 'compressors' / c1*c10 /
jii(j,ii,i) 'connections'
/ c1.s0.s1, c2.s1.s2, c3.s2.s3, c4.s3.(s4,s8), c5.s4.s5
c6.s5.s6, c7.s6.s7, c8.s8.s9, c9.s9.s10, c10.s10.s11 /
qi(ii,i) 'pipeline flow mapping'
cj(j,ii) 'compressor inflow mapping'
ji(j,ii) 'compressor pipeline flow mapping'
is(ii) 'all segments less terminal segments';
cj(j,ii) = sum(i, jii(j,ii,i));
qi(ii,i) = sum(j, jii(j,ii,i));
ji(j,i) = sum(ii, jii(j,ii,i));
is(ii) = sum(j, cj (j,ii));
Parameter
k 'suction condition' / 1.26 /
z 'compressibility factor' / 0.88 /
t 'suction temperature' / 520 /
cvar 'compressor annual capital variable cost ($ per hp)' / 70 /
cfix 'compressor annual fixed cost ($)' / 1e4 /
pcost 'pipe capital cost ($ per inch per mile)' / 870 /
ocost 'operating cost ($ per hp)' / 8 /;
Variable
l(i) 'pipe length'
Pd(i) 'discharge pressure (psi)'
Ps(ii) 'suction pressure (psi)'
D(i) 'diameter (inches)'
Q(ii) 'flow (MMCFD)'
CR(j) 'compression ratio'
W(j) 'work (hp)'
b(j) 'compressor decision'
cl 'pipe capital cost'
cc 'compressor capital cost'
oc 'operating cost'
cost;
Binary Variable b;
Positive Variable w;
Equation
pdrop(i) 'pressure drop in lines'
ltwo 'line length two'
lthree 'line length three'
qdef(i) 'flow definitions'
qbal(ii) 'flow balances and gas consumption'
wdef(j,i) 'work definition'
crdef(j,ii,i) 'compression definition'
cdef(j)
cldef 'cost of pipes'
ccdef 'cost of compressors'
ocdef 'operating cost'
obj 'total cost';
pdrop(i).. pd(i) =g= ps(i);
ltwo.. sum(i1, l(i1)) + sum(i2, l(i2)) =e= 175;
lthree.. sum(i1, l(i1)) + sum(i3, l(i3)) =e= 200;
qdef(i).. l(i) =e= sqr(871*1e-6)*d(i)**(16/3)*(sqr(pd(i)) - sqr(ps(i)))/sqr(q(i));
crdef(jii(j,ii,i)).. cr(j)*ps(ii) =e= pd(i);
cdef(j).. cr(j) =l= 1 + (cr.up(j) - 1)*b(j);
qbal(is).. q(is) =e= sum(qi(is,i), q(i)) + sum(cj(j,is), 0.005*q(is)*b(j));
wdef(ji(j,i)).. w(j) =e= 0.08531*q(i)*k/(k - 1)*t*(cr(j)**(z*(k - 1)/k) - 1);
cldef.. cl =e= sum(i, pcost*l(i)*d(i));
ccdef.. cc =e= sum(j, cvar*w(j) + cfix*b(j));
ocdef.. oc =e= sum(j, ocost*w(j));
obj.. cost =e= cl + cc + oc;
Model gasnet / all /;
* simple bounds
pd.lo(i) = 200;
pd.up(i) = 1000;
ps.lo(i) = 200;
ps.up(i) = 1000;
L.lo(i) = 2;
L.up(i) = 200;
D.lo(i) = 4;
D.up(i1) = 36;
D.up(i2) = 18;
D.up(i3) = 18;
cr.lo(j) = 1;
cr.up(j) = 2;
q.lo(i) = 200;
q.up(i) = 600;
ps.fx('s0') = 500; // fix starting pressure at line one
ps.fx('s7') = 600; // fix end pressure at line two
ps.fx('s11') = 300; // fix end pressure at line three
q.fx('s0') = 600; // fix inflow at line one
l.l(i) = 20;
d.l(i) = 18;
solve gasnet minimizing cost using minlp;
Parameter
prep 'pipe report'
crep 'compressor report';
prep(i,'Discharge') = pd.l(i);
prep(i,'Suction') = ps.l(i);
prep(i,'Diameter') = d.l(i);
prep(i,'Length') = l.l(i);
prep(i,'Flow') = q.l(i);
crep(j,'Ratio')$b.l(j) = cr.l(j);
crep(j,'Work' )$b.l(j) = w.l(j);
display prep, crep;