SLIDE 1
Oil, Gas and Chemical CFD Conference November 4-5, 2014
DE DEM s solu lutions ns f for O Oil & l & Ga Gas, , and nd C Che hemi mical i l ind ndustries
Oleh eh B Baran
SLIDE 2 DE DEM f for mo modeli ling ng r rock d k drilli lling ng
– Breaking rock challenges – Relevant capabilities in STAR-CCM+ – Example without coupling to fluid flow – Using overset mesh to model drill-bit motion
DE DEM f for mo modeli ling ng f flo low o
soli lids i in f n flu luidized b beds
– Coarse-grain model in STAR-CCM+ – Industrial scale fluidized bed example
- Simulation results for large particle size distribution
Summa mmary y
Outli line ne
2
SLIDE 3 Rock C k Cutting ng C Comple lexity y
Rock c k cutting ng: :
– Complex non-equilibrium and non- steady-state processes – Wide range of length-scales
- From grain scale
- To bore-hole / reservoir dimensions
– Wide range of time-scales
- From sound waves period in solids
- To hours of advancing drill-bit
through inhomogeneous rock
Can nu n nume merical mo l modeli ling ng he help lp?
– In improving drill-bit design – In optimizing operation parameters (rpm, ROP, WOB) – Reduce bit balling
3
SLIDE 4 Modeli ling ng R Rock u k using ng DE DEM
DE DEM mo models ls i ind ndividuals ls g grains ns i in n rock k
– Accurate grain scale physics
- Resolution of grains-cutter contacts
- Can reproduce removing cuttings
– Limited to smaller length scales and timescales
Model i l inputs
– Bit Design
- Nozzle selection
- Teeth configuration, etc
– Operation parameters
- Weight on Bit (or ROP), RPM…
– Rock properties
Observable les
– Rate of penetration (or WOB) – Torque – Cuttings attached to drill bit – Stand pipe pressure
Model c l cha halle lleng nges
– Simulation time – Far Boundaries – Calibrating model of rock – Simulating flow of drilling fluid in borehole – Reproducing bit balling – Reproducing realistic cutting flows
4
SLIDE 5 The he P Paralle llel B l Bond nds mo model i l int ntroduces a attractive int nter-p
le f forces t to t the he p particle le s sys ystem m Model u l uses t the he c conc ncept o
a ma massle less b bar conne nnecting ng a a p pair o
bond nded p particle les The he b bar(bond nd) c can t n trans nsmi mit f force a and nd t torque between p n particle les a and nd i it i is a als lso s subje ject t to b breaki king ng und nder lo load
– The stress limit values are calculated based on beam theory
Referenc nce: : Potyond ndy, D.O , D.O, a , and nd Cund ndall ll, P , P.A .A. 2 . 2004. . “A b bond nded-p
le mo model f l for r rock” k”, Int , Int. J . J. R . Rock k Mecha hani nics & & M Mini ning ng S Scienc nces 4 41 p
. 1329–1364.
DE DEM P Paralle llel B l Bond nds M Model i l in S n STAR-C
SLIDE 6 Dr Drilli lling ng e example le s set-u
6
SLIDE 7
Result lts
7
SLIDE 8 ~75000 p particle les w with h Ga Gaussian s n size d distribution n whi hich h Particle les s settle led a and nd b bond nded with a h about 1 180000 b bond nds Bond nd s streng ngth d h distributed according ng t to Ga Gaussian n distribution ( n (with me h mean v n valu lue
bond nd s streng ngth = h = 1 1% o
Young ng’s ’s mo modulu lus). .
Moni nitoring ng a amo mount nt o
broken b n bond nds
8
SLIDE 9
Cont ntact ne network k
9
SLIDE 10
Cont ntact c colo lored b by ‘B y ‘Bond nd S State’ ’
10
SLIDE 11 Modelli lling ng d drilli lling ng f flu luid
Possible le i in la n latest v version 9 n 9.0 .06 because o
compatibili lity o y of DE DEM with O h Overset M Mesh h
11
SLIDE 12 Result lts w with o h overset me mesh h
Rock i k is p perme meable le w with v h void fraction = n =0.4 .4 Solu lution f n for d drilli lling ng f flu luid f flo low w was
ned u using ng 2 2-w
y coupli ling ng mo model l Jet f flo low f form no m nozzle les r result lts i in n la large d drag f forces o
n bond nded g grains ns
12
SLIDE 13
After d drilli lling ng – – b before C Che hemi mical p l processing ng
13
SLIDE 14 Ind Industrial S l Scale le F Flu luidized B Bed S Study y
14 3 m d=0.6 m 0.4 m d=0.1 m 0.46 m
air inlet distributor Mesh size 20 mm
SLIDE 15
Amo mount nt a and nd s size d distribution o n of p particle les
15
Fines: Mass:~ 108 kg Diameter: ~500 microns Count: ~716,000,000 Coarse particles: Mass:~ 108 kg Diameter: ~1mm Count: ~80,000,000
SLIDE 16 DE DEM p parcel r l represent nts s some me nu numb mber o
ident ntical u l unr nresolv lved p particle les 𝒆 𝒆↓𝒒𝒃𝒔𝒅𝒇𝒎 =𝒎 ¡𝒆 𝒎≥𝟐 𝒎 𝒎↑𝟒 -nu
mber o
particle les i in p n parcel l Flu luid-p
le i int nteraction ( n (drag, li , lift e etc.) .) a are c calc lcula lated f for a a represent ntative p particle le a and nd a appli lied t to t the he e ent ntire p parcel l
– while the contact dynamics are calculated on the parcel scale
Coarse Gr Grain P n Particle le
16
part rticle icles s parce rcel l
Faster DEM computing time
SLIDE 17
Example le f for Gi Gidaspow d drag f force c calc lcula lation n
𝑫↓𝒆 𝑫↓𝒆 ={█□𝟓/𝟒 (𝟐𝟔𝟏 𝟐𝟔𝟏(𝟐−𝜷↓𝒒 𝜷↓𝒒 )/𝜷↓𝒒 𝜷↓𝒒 𝑺𝒇↓𝒒 𝒇↓𝒒 +𝟐.𝟖𝟔 𝟖𝟔); ¡ ¡ ¡ ¡ ¡ ¡if ¡ ¡ ¡𝜷↓𝒒 𝜷↓𝒒 <𝜷↓𝒏𝒋𝒐 𝜷↓𝒏𝒋𝒐 ;𝐅𝐬𝐡 𝐅𝐬𝐡𝐯𝐨 ¡ 𝐟𝐫𝐯 𝐫𝐯𝐛𝐮𝐣𝐩𝐨 𝐩𝐨 ¡𝐜𝐛 𝐜𝐛𝐭𝐟𝐞@𝟑𝟓 𝟑𝟓+𝟒.𝟕𝑺𝒇↓𝒒 𝒇↓𝒒↑𝟏.𝟕𝟗𝟖 𝟕𝟗𝟖 /𝑺𝒇↓𝒒 𝒇↓𝒒 𝜷↓𝒒 𝜷↓𝒒↑−𝟒.𝟕𝟔 𝟕𝟔 ; ¡ ¡ ¡ ¡ ¡ ¡if ¡ ¡ ¡ ¡𝜷↓𝒒 𝜷↓𝒒 ≥ 𝜷↓𝒏𝒋𝒐 𝜷↓𝒏𝒋𝒐 ; ¡ ¡𝐗𝐟 𝐗𝐟𝐨 ¡𝐙𝐯 𝐙𝐯 ¡𝐧𝐩𝐞 𝐩𝐞𝐟𝐦 ¡ ¡ ¡ ¡ ¡ ¡ ¡ – Here 𝛽↓𝑞 is solid void fraction, 𝛽↓𝑛𝑗𝑜 is the cutoff void fraction (=0.8), 𝑆𝑓↓𝑞 is the particle Reynolds number, 𝑒 is particle diameter 𝐺↓𝑒𝑠𝑏 (𝑞𝑏𝑠𝑑𝑓𝑚)=𝑚↑3 𝟐/𝟑 𝝇↓𝒈 𝝇↓𝒈 𝒘↑ 𝒘↑𝟑 𝑫↓𝒆 𝑫↓𝒆 𝑩↓ 𝑩↓𝒆 - this drag force is applied to parcel containing 𝑚↑3 particles, as a result: Acceleration, velocity and displacement of parcels due to scaled drag is similar to acceleration, velocity and displacement and of fine particles due to original unscaled drag force
Coarse Gr Grain De n Details ls
17
SLIDE 18
Parcel S l Size d distribution n
18 Fine: 𝒎=𝟐𝟗
𝟐𝟗; ¡ ¡ ¡𝒎↑ 𝒎↑𝟒 =𝟔,𝟗𝟒𝟑 𝟗𝟒𝟑
𝒆
𝒆↓𝒒𝒃𝒔𝒅𝒇𝒎 = 9 mm; 𝑶↓ 𝑶↓𝒒𝒃𝒔𝒅𝒇 𝒒𝒃𝒔𝒅𝒇𝒎 𝒎 = 109,000
Coarse: 𝒎=𝟐𝟗
𝟐𝟗; ¡ ¡ ¡𝒎↑ 𝒎↑𝟒 =𝟔,𝟗𝟒𝟑 𝟗𝟒𝟑
𝒆
𝒆↓𝒒𝒃𝒔𝒅𝒇𝒎 = 18 mm; 𝑶↓ 𝑶↓𝒒𝒃𝒔𝒅𝒇 𝒒𝒃𝒔𝒅𝒇𝒎 𝒎 = 14,000
Fine: 𝒎=𝟑𝟓
𝟑𝟓; ¡ ¡ ¡𝒎↑ 𝒎↑𝟒 =𝟐𝟒 𝟐𝟒,𝟗𝟑𝟓 𝟗𝟑𝟓
𝒆
𝒆↓𝒒𝒃𝒔𝒅𝒇𝒎 = 12 mm; 𝑶↓ 𝑶↓𝒒𝒃𝒔𝒅𝒇 𝒒𝒃𝒔𝒅𝒇𝒎 𝒎 = 46,000
Coarse: 𝒎=𝟐𝟑
𝟐𝟑; ¡ ¡ ¡𝒎↑ 𝒎↑𝟒 =𝟐,𝟖𝟑𝟗 𝟖𝟑𝟗
𝒆
𝒆↓𝒒𝒃𝒔𝒅𝒇𝒎 = 12 mm; 𝑶↓ 𝑶↓𝒒𝒃𝒔𝒅𝒇 𝒒𝒃𝒔𝒅𝒇𝒎 𝒎 = 46,000
SLIDE 19 W We d define ne t the he me mean s n size o
particle les 𝑒 𝑒
– Equivalent mono-disperse system provides the same total surface area
𝑒 𝑒 =∑ ∑𝑗=1 =1↑𝑂 𝑂↓𝑞𝑏𝑠𝑑𝑓 𝑑𝑓𝑚𝑡 ▒𝑚 𝑚↓𝑗↑3 𝑒 𝑒↓𝑗↑3 /∑𝑗 /∑𝑗=1 =1↑𝑂 𝑂↓𝑞𝑏𝑠𝑑𝑓 𝑑𝑓𝑚𝑡 ▒𝑚 𝑚↓𝑗↑3 𝑒 𝑒↓𝑗↑2 The he f frictiona nal p l pressure d drop 𝛼𝑄 𝛼𝑄 f for b bed w with s h size d distribution n f for b bed w with s h size d distribution n −𝛼 𝛼𝑄/𝑀 =150 =150( (1− 1−𝛽 𝛽 ↓𝑞 )↑ )↑2 /𝛽 𝛽 ↓𝑞 ↑3 𝜈 𝜈↓𝑔 𝑣/𝑒 𝑒 ↑2 +1.75 +1.75( (1− 1−𝛽 𝛽 ↓𝑞 )/ )/𝛽 𝛽 ↓𝑞 ↑3 𝜍↓𝑔 𝑣 𝑣↑2 /𝑒 𝑒
Size Di Distribution C n Correction t n to P Pressure Dr Drop ( (The heory) y)
19
SLIDE 20 The he f flu luidization v n velo locity y 𝑣 𝑣↓𝑛𝑔 e estima mate f for mi mixed p particle les s sizes 𝑣 𝑣↓𝑛𝑔 =(√ (√(𝐿 (𝐿↓2 / /2𝐿 𝐿↓1 )↑ )↑2 +𝐵 𝐵𝑠/𝐿 𝐿↓1 −𝐿 𝐿↓2 /𝐿 𝐿↓1 ) ) ¡ ¡𝜈↓𝑔 /𝜍 𝜍↓𝑔 𝑒 𝑒 ¡ ¡ 𝐿 𝐿↓1 =1.75 1.75/𝛽 𝛽 ↓𝑞↑3 ; ¡ ; ¡ ¡ ¡ ¡ ¡ ¡ ¡𝐿 𝐿↓2 =150 =1501− 1−𝛽 𝛽 ↓𝑞 /𝛽 𝛽↓𝑞↑3 ; ¡ ; ¡ ¡ ¡ ¡ ¡ ¡ ¡𝐵𝑠 𝐵𝑠=𝜍 𝜍↓𝑔 (𝜍 (𝜍↓𝑡 −𝜍 𝜍↓𝑔 ) )/ 𝜈↓𝑔 𝜈↓𝑔↑2 𝑒 𝑒 ↑3 For o
test c case 𝑣 𝑣↓𝑛𝑔 =𝟏 =𝟏.𝟒𝟒 𝟒𝟒 ¡𝐧/𝐭 ( (used 𝛽 𝛽 ↓𝑞 =0.396 =0.396 o
ned f from m ana nalys lysis o
the he s simu mula lation r n result lts)
Flu luidization v n velo locity ( y (the heory) y)
20
SLIDE 21 Inlet velocity (m/s) 5 7 8 9 10 11 Superficial (m/s) 0.139 0.194 0.222 0.250 0.278 0.306 Inlet velocity (m/s) 12 13 15 20 30 50 Superficial (m/s) 0.333 0.361 0.417 0.556 0.833 1.39
Simu mula lation d n details ls
21
L(t)
Cells for averaging void fraction to get 𝛽
𝛽↓𝑞 (𝑢 (𝑢)
Height of bed 𝑀(𝑢
(𝑢)=2 ¡ =2 ¡〈𝑧 〈𝑧〉+ 〈𝑒 〈𝑒〉
is double of center of mass height
SLIDE 22
Result lts f for b bed he height ht a and nd v void f fraction n
22
SLIDE 23 Pressure Dr Drop i is c calc lcula lated a as s surface a average o
pressure o
n the he i inle nlet b bound ndary y
Result lts f for P Pressure Dr Drop
23
SLIDE 24 Time me-a
result lts
24
SLIDE 25
Flu luidization A n Ani nima mations ns d d9d18
25
0.25 m/s 0.31 m/s 0.33 m/s 0.36 m/s 0.56 m/s
SLIDE 26
Superficial = l = 1 1.3 .39 m/ m/s
Fr Freeboard
26
d9d18 d12d12
SLIDE 27
Segregations ns
27
d9d18 d12d12
SLIDE 28 Summa mmary y
Novel a l approach i h in u n using ng s scale le distribution f n for p particle les o
different nt s size o
better c cont ntrol l
the he s size o
the he s sma malle llest p parcel l in t n the he s simu mula lation, t n, total p l parcel l count nt a and nd, c , cons nsequent ntly ly, t , the he performa manc nce o
the he mo model l
28
SLIDE 29
Oil, Gas and Chemical CFD Conference November 4-5, 2014 Tha hank Y nk You!
