Holger Hermanns Holger Hermanns E le c tr ic Powe r Distr - - PowerPoint PPT Presentation

Holger Hermanns

Holger Hermanns

E le c tr ic Powe r Distr ibution Gr ids

Consume r s Pr

• duc e r

s Consume r s

nuc l c o a l

Pr

• duc e r

s

le a r g a s b io g a s

Micro CHP la ke s

Holger Hermanns

Stor age

Pr inc ipal E le c tr ic ity Mar ke t Conside r ations

e le c tric po we r de ma nd ne ve r e xc e e ds the po te ntia l o ffe r p

• pro duc e rs a re fully c o ntro lla b le

p y

• b a re ly a ny re g ula tio n o n the c o nsume r side

O it i i l

• pro duc e rs a re ha rdly c o ntro lla b le

Opposite pr inc iple s

pro duc e rs a re ha rdly c o ntro lla b le

• simple me c ha nisms to c o ntro l the c o nsume r side

e le c tric po we r c a n o nly b e

Holger Hermanns

e le c tric po we r c a n o nly b e c o nsume d if it is a va ila b le

E le c tr ic fying Challe nge

Ge r many

• g ive s prio rity to

re ne wa b le so urc e s re ne wa b le so urc e s

• re wa rds re ne wa b le e ne rg y

re wa rds re ne wa b le e ne rg y a b o ve ma rke t pric e

• dro ppe d nuc le a r e ne rg y

a fte r F ukushima inc ide nt Mo re c ha lle ng e s:

so urc e : http:/ / www wikipe dia o rg

Holger Hermanns

g Swe de n, UK , F ra nc e , ….

so urc e : http:/ / www.wikipe dia .o rg

E le c tr ic tifying Basic s

Guiding princ iple : Pr

• duc tion follows Consumption

Ba se Assumptio ns:

Co nsumptio n is a we ll-unde rsto o d sto c ha stic pro c e ss p p Pro duc tio n is de te rministic a nd fully c o ntro lla b le Ma ss e ffe c ts e nsure smo o th c o nsume r b e ha vio ur Grid sta te is o b se rva b le (fre q ue nc y vo lta g e ) Grid sta te is o b se rva b le (fre q ue nc y, vo lta g e )

T hus, g o o d pre dic tio ns a nd a b it o f fine tuning do the jo b .

Contr

• l e ne r

gy, c a n b e sub tra c te d

• r a dde d to the g rid. Mo stly pump sto ra g e .

I nde e d, this wa s the c a se fo r se ve ra l de c a de s.

g y p p g

Holger Hermanns

No w: T hing s c ha ng e ra pidly. Minds c ha ng e slo wly.

Holger Hermanns

So, what‘s the c halle nge ?

T he inte gr ation of r e ne wable e ne r gy R b l h h hi h l tilit d thi l tilit i t ll b l Re ne wa b le e ne rg y ha s muc h hig he r vo la tility a nd this vo la tility is unc o ntro lla b le . Pro duc tio n turns into a sto c ha stic pro c e ss, a s we ll! Ne e de d: inc re a se d pre dic tio n e ffo rts fo r g rid sta b iliza tio n.

Volatility may e xc e e d the available c ontr

• l e ne r

gy.

On Se pte mb e r 6 2010: On Se pte mb e r 6, 2010: Dra stic a lly mo re so la r po we r in the ne t tha n a nno unc e d the da y b e fo re Ge rma ny @ lunc htime : surplus o f 7 GW Co mple te ne g a tive c o ntro l e ne rg y e xha uste d (- 4.3 GW) E me rg e nc y re se rve impo rte d fro m ne ig hb o uring c o untrie s (- 2.8 MW)

Holger Hermanns

g y p g g ( )

Challe nge s for E c onomic E le c tr ic ity Usage ?

• Vo la tility e ffe c ts the ma rke t pric e fo r sho rt te rm e le c tric ity.
• Cha ng e s wo rklo a d c ha ra c te ristic s fo r tra ditio na l po we r pla nts

Cha ng e s wo rklo a d c ha ra c te ristic s fo r tra ditio na l po we r pla nts.

S

• me Co nc e rns:

What happe ns o nc e re ne wable e ne rgy pro duc tio n What happe ns o nc e re ne wable e ne rgy pro duc tio n is highe r than to tal c o nsumptio n? What pro duc tio n e ntitie s are ne e de d,

Holger Hermanns

p if all base c o nsumptio n is c o ve re d by re ne wable e ne rgy?

[H, Wie c hma nn: E F T A 2009]

How to make mone y with this?

Holger Hermanns

while ke e ping the ne twor k stable ?

What to c ontr

• l on the c onsume r

side ?

• L

ig ht b ulb s?

• I

ro ning ?

• PC?
• T

V?

• E

le c tric a l wa te r wa rming ?

• Clima te c o ntro l?
• Co o ling ?
• Air pre ssure a pplic a tio ns?
• Off-pe a k sto ra g e he a ting ?
• Ge o the rma l he a ting ?
• E

le c tric a l a nd hyb rid e le c tric a l ve hic le s?

Holger Hermanns

T he se g me nt o f e lastic c onsume r

s in Ge rma ny

is in the o rde r o f a fe w te ns o f GW.

Challe nge s for Stable E le c tr ic ity Usage ?

• Vo la tility a nd sync hro ny o f so la r po we r a ffe c ts g rid sta b ility

E WE inte rve ntio ns2009: <1 pe r we e k 2011: >1 pe r da y

• 75% o f a ll insta lla tio ns a re no n me a sure d mic ro -g e ne ra to rs

a re b a la nc e d o ut o nc e pe r ye a r.

• Sta rting 2000 re g ula tio ns a re b e ing a djuste d a t a n inc re a sing pa c e
• Sta rting 2000, re g ula tio ns a re b e ing a djuste d a t a n inc re a sing pa c e .
• L
• c a l distrib utio n g rid c o o rdina to rs thus fa r re po rte d a ve ra g e s
• witho ut c o nside ring up-to -da te we a the r c o nditio ns
• witho ut c o nside ring up-to -da te we a the r c o nditio ns

se a so na l diffe re nc e s, da y-nig ht diffe re nc e s

for bidde n as of Januar y 1, 2011

• T

a rg e t g ro wth 1.5 GW pe r ye a r a s o f 2009 -- the n10 GW Ac tua l g ro wth in 2011 is 7.5 GW -- no w 25 GW

y ,

Holger Hermanns

g

A c onc r e te pr

• ble m:

Distr ibute d Stabilisation Distr ibute d Stabilisation

T he pro ble m:

L e g a l b o die s rule d in 2007 tha t a so la r po we r L e g a l b o die s rule d in 2007 tha t a so la r po we r g e ne ra to r o b se rving the fre q ue nc y to o ve rsho o t 50 2 He rtz must shut o ff 50,2 He rtz, must shut o ff. Be c a use o f the sync hro nic ity o f o b se rva tio ns Be c a use o f the sync hro nic ity o f o b se rva tio ns, this le a ds to o sc illa tio ns, if no t b la c ko uts.

What we do :

We g e t inspire d b y ne two rk pro to c o ls We g e t inspire d b y ne two rk pro to c o ls.

Holger Hermanns

L ast mile distr ibute d c ontr

• l

“T

• o many mic ro ge ne rato rs

t ll t t li d t l“ to ro ll o ut c e ntralize d c o ntro l“

S lf t b ili ti b l Se lf sta b iliza tio n pro b le m b ut with a sha re d sta te : F re q ue nc y a nd vo lta g e Osc illa tio ns a re a pro b le m. Curre nt sta te o f le g isla tio n: o ff a t 50.2 Hz C t t t f i i

Holger Hermanns

Curre nt sta te o f e ng ine e ring : L ine a r de c re a se b e twe e n 50.1 a nd 52.2

What the r ule s say

Holger Hermanns

Re a c tio n de la y 0, 10 o r 20 se c o nds.

Analysis: Mode l Che c king

Requirements Real System Requirements Real System

Modelling Formalising

System Model

possible behaviour

Requirement Specification

allowed behaviour

Model Checker Model Checker

Next Requirement Requirement

Holger Hermanns

No

Counterexample

YES

done

D

e Pr

A

• v

e x

Add

wi dis

T his is re a l: www.mo de stc he c ke r.ne t

Holger Hermanns

What the VDE sugge sts

Holger Hermanns

What we do

Study po pula tio ns o f so la r pro duc e rs in a la st mile so la r pro duc e rs in a la st mile Co ntro lle rs

• n o ff

with o r witho ut re a c tio n time

• n-o ff

VDE pro b a b ilistiic

with o r witho ut e xpo ne ntia l b a c ko ff

pro b a b ilistiic with dyna mic size o f die

Sync hro no us, a sync hro no us, drifting Sto c ha stic b a c kg ro und lo a d sc e na rio s

Co nsta nt Pro file s Ra ndo m wa lk

Holger Hermanns

Ma rko v mo dula te d Be rno ulli pro c e sss

Additive -inc r e ase , multiplic ative de c r e ase

Use d in:T CP Go a l: Use ma xima l b a ndwidth, b ut sha re it fa irly , y I de a : I nc re a se use in sma ll, a dditive ste ps De c re a se o n b a d e ve nt b y a fa c to r < 1 De c re a se o n b a d e ve nt b y a fa c to r < 1

fre q ue nc y 50.2 Hz 50.0 Hz time

Holger Hermanns

What we ar e thinking of

AI MD: Additive inc re a se , multiplic a tive de c re a se

Holger Hermanns

F r e que nc y-de pe nde nt pr

• babilistic switc hing

pr

• babilistic switc hing

Use d in:I E E E802.11e Use d in:I E E E 802.11e Go a l: Ada pt to syste m sta te , b ut ra ndo mise d I de a : S itc h o n o r o ff ith pro b a b ilit I de a : Switc h o n o r o ff with pro b a b ility de pe nde nt o n o b se rve d fre q ue nc y

pro b a b ility 1

• ff
• n

50 0 Hz fre q ue nc y 50 2 Hz 50 4 Hz

• ff

Holger Hermanns

50.0 Hz 50.2 Hz 50.4 Hz

Mor e of this kind

F re q ue nc y-b a se d pro b a b ilistic switc hing

Holger Hermanns

So?

Goodput

Re a c tio n de la y 10 se c o nds

Holger Hermanns

Availability

So?

Goodput

Re a c tio n de la y 10 se c o nds

Holger Hermanns

Availability

So?

Goodput

Re a c tio n de la y 20 se c o nds

Holger Hermanns

Availability

Our c ur r e nt favor ite :

F re q ue nc y b a se d pro b a b ilistic switc hing plus e xpo ne ntia l b a c ko ff

Holger Hermanns

E xpone ntial bac koff

Use d in: E the rne t Go a l: Me dia te a c c e ss to sha re d me dium, de c e ntra lise d , I d T t d I de a : T ry to se nd. Co llisio n? Wa it time g ive n b y 12-side d die ro ll. T ry to se nd T ry to se nd. Co llisio n? Wa it time g ive n b y 14-side d die ro ll. T ry to se nd. Co llisio n? Wa it time g ive n b y 18-side d die ro ll. T ry to se nd. C lli i ? W it ti i b 16 id d di ll Co llisio n? Wa it time g ive n b y 16-side d die ro ll. …

Holger Hermanns

a djust wa it time to numb e r o f c o mpe tito rs

Whe r e do the se numbe r s c ome fr

• m?

32 pho to vo lta ic mic ro g e ne ra to rs. De c ide upo n o b se rve d fre q ue nc y e ve ry 20 se c . T ime po ints initia lly pic ke d unifo rmly. L ine a r impa c t o n fre q ue nc y. Re a c tio n de la y is 0, 10 o r 20 se c o nds. Ba c kg ro und fre q ue nc y g e ne ra te d b y ra ndo m wa lk. All mo de lle d in Mo de st. 10000 simula tio n runs o f mo de s disc re te e ve nt simula to r.

Holger Hermanns

F air ne ss

Ma x/ min/ a ve ra g e

• utput

pe r g e ne ra to r:

Holger Hermanns

What r e mains to be done ?

A lo t. T e a m up with de sig n spa c e e xplo ra to rs de sig n spa c e e xplo ra to rs ma c hine le a rne rs do ma in e xpe rts de c isio n b o die s pa te nt fa na tic s Co nside r vo lta g e inste a d o f fre q ue nc y

spa tia l la yo ut c a n b e c o me impo rta nt (de pe nde nt o n se tup) p y p ( p p)

De ve lo p g o o d a b stra c tio ns

Holger Hermanns

I nc re a se a wa re ne ss

Conc lusion

Pho to vo lta ic mic ro g e ne ra tio n c re a te s ne w c ha lle ng e s. F

• rma l mo de ls he lp e va lua te ne w a ppro a c he s so undly.

So me o f the ne w c o ntro l stra te g ie s wo rk re a lly we ll

Bo nus: No privac y c o nc e rns!

So me o f the ne w c o ntro l stra te g ie s wo rk re a lly we ll.

WSC 2012

Holger Hermanns

WSC 2012

ISoL A 2012

Holger Hermanns

A br

• ade r

pe r spe c tive

On the lo ng run, th id t b ilit b l ill b the po we r g rid sta b ility pro b le m will b e a ma jo r c o nc e rn o n a ll la ye rs f th id hi h

• f the g rid hie ra rc hy

Wh t i d d i id t ti Wha t is ne e de d, is a g rid ma na g me nt o pe ra ting

• highly loc al,
• highly automatic ,
• highly de c e ntr

alize d, and

• highly fle xible

Holger Hermanns

Ne w ope r ational pr inc iple s for powe r gr ids

Gr id ne utr ality: T

he re is no disc rimina tio n in the wa y the g rid sha re s its c a pa c ity a mo ng its use rs.

Gr id fair ne ss: I

de a lly, the g rid is fa ir in the se nse tha t if n use rs a re sha ring a g rid pa th, the n o n a ve ra g e e a c h use r c a n use b t 1/ th f th it a b o ut 1/

n-th o f the c a pa c ity.

Sc alability:

Distrib ute d, de c e ntra lize d c o ntro l is a prime me a ns to a ssure sc a la b ility, to g e the r with hie ra rc hy.

Pr ivac y: E

nd use r priva c y is pro te c te d b y de c e ntra lize d

Pr ivac y: E

nd use r priva c y is pro te c te d b y de c e ntra lize d de c isio n ma king b a se d o n pub lic g lo b a l info rma tio n a nd priva te lo c a l info rma tio n priva te lo c a l info rma tio n.

Inte llige nt e dge s, dumb c or e: I

nte llig e nc e re side s in the

Holger Hermanns

e dg e s o f the g rid, i.e . is e mb e dde d into the e nd use r a pplia nc e s. T he c o re o f the g rid is b a re ly sma rt.

Ne w ope r ational pr inc iple s ar e old!

Gr id ne utr ality: T

he re is no disc rimina tio n in the wa y the g rid sha re s its c a pa c ity a mo ng its use rs.

Gr id fair ne ss: I

de a lly, the g rid is fa ir in the se nse tha t if n use rs a re sha ring a g rid pa th, the n o n a ve ra g e e a c h use r c a n use b t 1/ th f th it a b o ut 1/

n-th o f the c a pa c ity.

Sc alability:

Distrib ute d, de c e ntra lize d c o ntro l is a prime me a ns to a ssure sc a la b ility, to g e the r with hie ra rc hy.

Pr ivac y: E

nd use r priva c y is pro te c te d b y de c e ntra lize d

Pr ivac y: E

nd use r priva c y is pro te c te d b y de c e ntra lize d de c isio n ma king b a se d o n pub lic g lo b a l info rma tio n a nd priva te lo c a l info rma tio n priva te lo c a l info rma tio n.

Inte llige nt e dge s, dumb c or e: I

nte llig e nc e re side s in the

Holger Hermanns

e dg e s o f the g rid, i.e . is e mb e dde d into the e nd use r a pplia nc e s. T he c o re o f the g rid is b a re ly sma rt.

An inte r ne t analogy!

Gr id ne utr ality: T

he re is no disc rimina tio n in the wa y the g rid sha re s its c a pa c ity a mo ng its use rs.

Gr id fair ne ss: I

de a lly, the g rid is fa ir in the se nse tha t if n use rs a re sha ring a g rid pa th, the n o n a ve ra g e e a c h use r c a n use b t 1/ th f th it a b o ut 1/

n-th o f the c a pa c ity.

Sc alability:

Distrib ute d, de c e ntra lize d c o ntro l is a prime me a ns to a ssure sc a la b ility, to g e the r with hie ra rc hy.

Pr ivac y: E

nd use r priva c y is pro te c te d b y de c e ntra lize d

Pr ivac y: E

nd use r priva c y is pro te c te d b y de c e ntra lize d de c isio n ma king b a se d o n pub lic g lo b a l info rma tio n a nd priva te lo c a l info rma tio n priva te lo c a l info rma tio n.

Inte llige nt e dge s, dumb c or e: I

nte llig e nc e re side s in the

Holger Hermanns

e dg e s o f the g rid, i.e . is e mb e dde d into the e nd use r a pplia nc e s. T he c o re o f the g rid is b a re ly sma rt.

T ur ning the powe r gr id into a powe r we b

T his is a vision.

Caution:

T he pro to c o ls a nd te c hniq ue s driving the we b ha ve b e e n de ve lo pe d driving the we b ha ve b e e n de ve lo pe d a t a time whe n no o ne c a re d.

T his time things ar e diffe r e nt: g

T he tra nsfo rma tio n must b e do ne while o ur so c ie ty re lie s o n it

Holger Hermanns

while o ur so c ie ty re lie s o n it if it is to b e do ne a t a ll.

Holger Hermanns