r 1 CONTROL AND MAINTENANCE PROBLEMS DURING OPERATI ON K. E. Breyayer Internal Meoranu M Report No. 277 August 1961 Project M Stanfor University Stanfor, California
TABLE OF CONTENTS Page. 1. Scopeof stuy.... 1 2. Maintenance an reliability... 1 3. Maintenance proble ung operation.... 1 4. Failures an their effect on operation.... 2 a. Paraeter changes... 2 b. Multiple-source failures.... 2 c. Single-source failures... 2. Failure rates an their effect on bea-on tie... 4 6. Repair of single-source failures... 4 a. Access tie... 4 b. Repair tie... 4 c. Service restoration tie... 4 7. Repair of ultiple-source failures... 6 8. Bea control requireents... 7 a. Klystron bea voltage... 7 b. Accelerator teperature... 8 c. Phase shift in the ve syste... 8. Suary of energy an phase vaations ue to paraeter changes... 8 9. Energy recovery schees... 9 a. Klystron bea-voltage ajustent.... 9 b. Klystron ve-level ajustent... 12 c. Klystron stanby... 12. Bea-loaing effect... 13 e. Pair-wise ephasing... 13 1. Bef copason an rating of the schees... 14 a. Klystron bea-voltage control... 14 b. Pair-wise ephasing... 14 c. Bea-loaing effects... 1 11. Suary... 1 T
1. SCOPE OF s'ru This stuy covers the probles associate with aintaining the analyse bea rrent of the two-ile accelerator within specifie energy an intensity liits ung the expeental tie. The effects of vaous failures on achine operation are analyse in etail an proceures are propose that ay be use to restore service. 2# MAINTENANCE AND IiELIABILITY Maintenance efforts norally are irecte towar a partilar ob- jective. The viewpoint taen here for this objective is the continuance of the analyse bea rrent within the esire specifie liits ung the expeental tie. The aintenance policy aopte shoul be irecte towar attaining this objective. 3. MAINTENANCE PROBLEM DURING OPERATION The objective, as entione above, is the continuance of the ana- lyse bea rrent. Partial or coplete loss of this rrent is coni- tione by two states the accelerator can assue in case of coponent failure, as follows. a. In the first state the bea is shut off in the accelerator an there is no analyse rrent. This is cause by a failure in a subsyste which contbutes a vital paraeter, e.g., the failure of the electron gun, the frequency generator, etc. Such failures will be calle single-source failures. Machine or personnel protective arrangeents (the interloc systes) that turn off the bea are also classifie as single - source failures e b. In the secon state the electron bea still exists in the accelerator but a change in the accelerator operating paraeters has ocrre, resulting in partial or coplete loss of the analyse bea. A change in raiofrequency power or rf phase usually causes this effect. Many inepenent subsystes contbute to the rf output, hence loss of a single subsyste prouces a change in the integrate rf power but oes not prouce a total loss of power. Such failures will be calle ultiple-source failures. The lystron power aplifiers an the associate supporting subsystes, such as water, vau, tgger, an ve power, contbute to ultiple-source failures. Phase an energy vaations also can be cause by vaations of the operating paraeters -1-
1 ung o2erhbioc. K-i potential failures can be classifie either as paraeter. ult ipie-source failures, or single-source failures. The probleni ientify the unerl-ying cause in case of aalyseci rrent failure fin appropate restoration proce6ures. 4. 3AlLX3ES AND THEIR EFFECT ON OPERA'TION a. Paraeter CiaFzes?'he paraeters affecng tne establishe bea an their relative infiinence are analyse in M Report No. 2.l change s is to The ost significant of an to these paraeters are lystror? bea voltage, operating teperature of the acceleret,or, ar, phase shipt in the rf systp. 'Yne vaations ue to paraeter changes are expecte to be "slow," i.e., in the tie range of several secorts to inutes or hours. The general efpect will be an analyse rrent that cont iniiously vaes in intensity. b. Multiple-Source Failures 1Ph~it iple-source failures have one coon charactestic : They change the scaation of the rt power in the accelerator, an hence the avail?hle Sea energy, by an aount that epens on the nuber OP sources isable The source of rf power (the lystron power aplifier) epens on a nuber of subsystes for successful operation. The subsystes consist of u'oer of utually inepenent uni-cs, each serving several lystrons. 'The sc?.llest power loss is inrre when a single lystron or oulator Iails. The increasing orer of losses epens on the subsyst,es, as +ilbu:atc beiow (see Table I) c'. Sinc'Le-Source Failures The coon charactestic ot single-source failures is the coplete loss of tea rrent in '<he accelerator. IThis can be cailse eit,lier by the JaiLure of a subsyste (e.g., the iiijector) which irectly affects LIP trsnsrnitte be or by protective shutown of an interloc cit when a hazar is creare. TIX co-plete loss OP rr power' an/or eiectron Toearn pulses can ocr. Ei1,hei. ay be catise by a failure 1 the I'r.eyuency or tgger generator. - 'K. E. Dreyayer, "Probles in 8ea Operation," M Report Do. 2q Pro;ject 14, SLanfor University, Stanfor, CaliIFoii June 196ie
x 8 4-1 $ R n v M A - x 8 * -$ w x rl R I PI -P Q t - - 3 -
The latter also ay be cause by a failure of either the injector or the bea steeng syste. 3. FAILURE RATES AND TEElR EFFEC'I 9 BEAM-ON TIME A table (Table 11) has been prepare, which Pollows the above classi- fications, showing average ailure rztes an estivate repair rates. rates giver! in this table shouicl not oe interprete as preicte values. Rather, the ata are eant to give a flrst orer estiate of the relative influence of' the vaous types of failures upon operation. Soe of the basic sssunptions use in the estiates are note in the footnotes to Table 11, 6. REPAIR OF SING~,-SOURCE FAILUFES By efinition, a single-source failure interrupts bea operation until service of the faile unit is restore Repair can be achieve in a nuniber of ways; e.g., the faulty coponent can be replace, stanby units can be use or efforts can be irecte towar preventing the failure fro ocrrlg in the first place. The foliowing ctea can be use to classify the expecte single- source failures with respect to the repair or EBintenance approach, proceures selecte can be calle a policy u; The Any This policy shoul be opti- i.e., it shoul iniize the outage tie, given the cost of alnte- nace, or the iniu cost policy shoui be foun, given tne pessible outage tie. a Access Tie W h ~ n no raiation hazar exists, the equipent In question will- be accessible ieiately, allowing only tie to get there. However, in case of raiation hazars, substantial elays ay exist. All. equipent outsie the acceierator housing ana switchyar norally will be iei- ately accessible. Insie the aceeleratar tunnel high raiatior, levels will exist an long elays of tne orer of' ays to wees ay result. ----- b- Repair Tie This is the tie fro the beginning of the actual repair tc the en. It inclues cietening the loclztio of tne exact, troubie spo-t in the fa: Lea equipent an the replaceent or ajustent of the faulty coponent. c Service --- RestoratSon ----- Tie This is the tie fro the en of the repair, when the faile equip- ent is restore to operatlo, to the resuption of bea opc:ration. - 4 - For
H H! R Ln aa rn '& H u f \D rn c-l a x $ u3 Pi co L n c- L n t- co in h 3 A- a ul I Ch A- f 4-1 2 2 tj G ul Pi PI Ln A- bd 2 Pi a 3 c tj f e 3 e 6) Pi r- e, L%. c 4-1,2 c Pi i? c vi " x 3 c tj ( i E h % P c i! 4-1 c c c, a 3 c a E v3 Pi c, 2 2 4 3 c : 3 a c. tj. 8 -$ 9 x Pi c x P.4. c 4-1 8 c, h tj c *A - c 4-1 c, Y Q tj h x 9 Pi 8 2 t, T
t exaple, this tie interval can be substantial in the case of vau failure. With respect to access tie an service restoration tie, two possible basic policies are recognize for iniizing outages: (1) avoiing the placeent of ctical coponents in areas having a long access tie, e.g., those in the accelerator housing or switchyar area; an (2) using preventive aintenance for equipent having a noral access tie but an extene service restoration tie. Soe of $he surface equipent locate in the lystron gallery will be of the latter type. The aintenance policy for the reaining surface equipent of the single-source type shoul be juge by its repair tie requireents. 7. REPAIR OF MULTIPLE-SOURCE FAILURES By efinition, ultiple-source failures lea to a change in rf power, but not necessaly to an integrate bea loss. The bea continues to exist in the accelerator but fails to reach the target because its energy has change. It is ischarge entirely on the slit plates an ay thus cause a raiation an heat proble. This situation etenes the viewpoint to be taen for restoration of operation. The overing consieration is the nee for quic eliination of the raiation an heat transfer hazar create, while aintaining a favorable bea-on tie ratio. The allowable tie for action with respect to eliinating the hazars is on the orer of uch less than one secon. There are two basic solutions available to eet these requireents. Either the bea can be turne off or the loss in rf power can be recovere quicly by suitable eans without turning off the accelerator. The two ost proising schees for restoration of bea operation have the following charactestics. a. Both eliinate the raiation hazar ieiately but with respect to bea-on tie soe ifferences exist. b. Turning off the bea will have an unfavorable effect on bea-on tie when the failure rate is high ue to the tie involve in getting bac to operation. c. The energy restoration schee is basically optiu, both to raiation eliination an bea-on tie, when esigne for an appropate autoatic proceure. -6-
As shown in Table 11, frequent events are associate with sall rf losses, ainly of the one-lystron type. frequency of the event ecreases, The opposite hols where the It appears that energy restoration shoul be attepte where rf power losses are frequent an sall. The bea turn-off" proceure shoul be use where the power losses are high an less frequent. In terns of the failure classifications, the bea turn-off proceure is equivalent to a single-source failurevin its effect, As such, it can be juge fro the isssion in Seetion 6. 8. BEAM @ONTI~OL REQ~IREMET~TS The pi-incipal operating paraeters contbuting to energy an phase vaations are the lystron bea voltage, the accelerator teperature, an the phase shift in the ve syste. 'These changes can prouce substantial vaations in the agnitue of the analyse rrent, as has been shown in M Report No, 2.2 Base on the relations escbe in that report, the following ata are eve, Klystron Bea Voltage Two effects result: the rf power output is change, an a phase shift is prouce. 1. Energy change ue to rf power change where.2 x is the assue tolerance on bea voltage. 2. Phase change ue to beau voltage change Cp - 1.2 X -2 x = - 2.3 x ra or 1.43' 3. Energy change ue to the phase vaation Report No, 23, op.cit. -7-
b. Accelerator Teperature The energy change associate with a change in teperature can be estiate fro the expression eve by Neal 3 V/V = 8=1 x lov3 (T)* ~V/V = 8.1 x x.3f = 1.11 x lov3 or.11% ' For. 7OoC, V/V = 8,1 x x.72 = 4 x lo-. 3 or.4$ c, Phase Shift in the Dve Syste A phase shift of ' is assue. This value ay be reache when all shifts in the ve syste ove in the sae irection. The axiu energy change associate with this shift can be assesse fro 2 -- V o2.9 v 2 2 --=-- -.4%. Suary of Energy an Phase Vaations Due to Paraeter Changes Assue Energy Phase Paraeter Tolerance Change Change Klystron bea voltage?r.2%.2 ra or 1.4' Accelerator teperature.37 C 2.7 c Dve syste phase f. p.4%.11%.4 ra or 2.'.4%.a ra or ' "'Project M Source Boo, It Californi Section II.B.3 (p. 22), July 196. Project M, Stanfor University, Stanfor, -8- T
9. ENERGY RECOVEBY SCHEMES Multiple-source failures an paraeter changes prouce energy an phase vaations that ust be offset in orer to aintain the intensity of the analyse rrent within specifie liits, The potential vaations prouce by paraeter changes are analyse above in Section 8. Each of the three paraeters consiere can prouce a axiu energy change of about.4%. The liit of the total change to be expecte is the linear su of the iniviual changes. This will. also be the liit of the control range for a schee use to offset these changes. A value of 1.% is assue. The tie oain of these changes, i.evj their rate of change, will be on the orer of secons to inutes. Multiple-source failures prouce iscrete changes in energy, The changes can tae place in units of 1 lystron, 8 lystrons, etc., as shown above in Table I. In Stage I a loss of one lystron represents a.4% loss of energy, an a loss of 8 lystrons represents a 3.2% loss of energy. Measures taen to recover such losses shoul be perfore in a short enough tie to iniize the heat an raiation probles they create. There is a nee for the continuous ajustent of energy ung bea operation through a potential range of about There is also a nee to recover energy lost, ue to losses in lystron power, on the orer of.4% an up. The latter requires fast action, The investigation of suitable energy recovery schees will be irecte towar arrangeents with an ajustent range capable of recoveng energy lost, as specifie above. Exaples of the suitable eans available are liste an issse below. a. Klystron bea-voltage ajustent b. Klystron ve-level ajustent e. Stanby lystrons. Bea-loaing effect e. Pair-wise ephasing a. Klystron Bea-Voltage Ajustent This schee pets all lystrons to operate at their noral level. In case of a loss, the applie voltage on all or soe of the operating -9-
tubes can be ajuste so that the rf power loss is recovere an a constant bea energy is aintaine. The ajustent of the bea voltage can be ae via the power supplies. A suitable control point is the so-calle "e-q' ing" cit of the oulator. A fast-acting control to aintain pulse-to-pulse voltage stability is provie there, Sall vaations in the e charging level can be obtaine via the e-q'ing reference voltage, There is one such control voltage for each 32 tubes of a coon power supply. The sie effects involve are twofol, A e voltage control of the lystron prouces a phase shift through the tube, an soe increase in power loss is inrre in the e-q'ing cit when the e charging level is reuce to obtain a suitable control argin. The phase shift generate can be offset in pnciple by the phase-copensating control. A etaile analysis of this proble will have to be ae to establish that the energy an phase control can wor inepenently of each other. The following exaple will illustrate the necessary bea-voltage ajustent, power increase, an energy control range. A general range of about l$, a iniu energy ajustent of O.l$, an, in aition, the proble of extening the range to about 376 will be consiere. The exaple eonstrates the use of one controlle power supply an failures within or without this unit. 1. Ajustent for a 1% an a Oel$ bea-energy control When N is the total nuber of lystrons an N1 the nuber to be ajuste, then the necessary increase of rf power in for an increase N1 in total energy V is given by 2* For the above case = Po (.1 x 24/32 + 1)* = 1.16% * Deve by K. Mallory. - 1 - I
NJ For a iniu energy ajustent of.1% the necessary power increase is.1 or lo%. P N V2/2, which hols when the lystron is operate uner optiu coni- tions. For s all changes in P1 the approxiation V/V 2/ P/P is use. The necessary increase in bea voltage is eve fro the relation Thus, a 1% ajustent in total energy requires a power increase of 16%, or a bea voltage increase of 6% in 32 lystrons. For.1% energy ajustent the power an voltage corrections are 1.% an.6% respect ively. 2. Ajustent when one lystron fails Two cases are consiere: the loss ocrs either insie or outsie the controlle unit. When N1 enotes the nuber of controlle lystrons an No the loss, then the increase in power in N1 is given by 2 * P =P 1 -Po + N1 For N1 = 32 the power increase in N1 an No = 1 an a loss outsie the controlling unit, is 6% an the bea-voltage increase is 2%. When the loss ocrs within the controlling unit N 1 power an voltage ajustents are 7% an 3% respectively. 3. Ajustent when 8 lystrons are lost = 31 an the The two situations uner the previous exaple are consiere. the loss is outsie the controlling unit N = 8 an N1 = 32. The power an voltage increase in is 7% an 2@ respectively. When Nl the loss ocrs within the controlle units N = 24 an N = 1 8. Power an voltage ajustent then becoe 78% an 23% respectively. If the bea voltage ajustent per lystron is liite to 16, the nuber of controlle units ust be increase. controlle power supplies, the siultaneous loss of 8 tubes can be co- When For N1 = 96, i.e., three pensate by the voltage ajustents of 7.2% an 7.6% respectively for insie or outsie losses. * Deve by K. Mallory. - 11 -
b. Klystron Dve-Level Ajustent Output vaations can be obtaine by control of the ve level. However, lystrons are norally operate uner "optiu" ve conitions. In such a case, the output power is rather insensitive to velevel vaations. In orer to obtain a suitable control argin, the input ve level has to be reuce below the optiu level. The rawbac of this ajustent is a reuction in operating efficiency of the lystron. There are also technical probles in ve-level control. Existing eans of control prouce substantial phase shifts. Fast controls are available at low power levels; however, for levels on the orer of ilowatts or higher, a echanical regulator is presently necessary. This woul lea to either great nubers of low-level controls for the fast ajustents or one or two high-level controls with slow reaction tie. This schee can be both continuous an fast in pnciple. However, it oes not appear very attractive because of the regulation probles an the sie effects entione above. c. Klystron Stanby In this syste a preassigne nuber of stanby units is set asie an the esire energy is establishe with the operating units. Whenever a loss ocrs, a stanby lystron is switche into operation ieiately as a replaceent for the isable unit. Stanby, petting quic operation, can be provie by reoving the ve, or by tgger blocing with the tube otherwise reay for operation, or by outphasing, i.e., tggeng between the ain pulses so that the rf generate has no influence on the electron bea. Tgger blocing reoves the c loa fro the lystron collector an reuces stress on the winows. The associate accelerator section will go through a teperature transient peo, when the tube is activate an rf is fe to the accelerator. The tube ay be operate up to 2 or 3 hours without sie effects on the cathoe when the bea voltage is reove. Outphasing, or tie-isplacing of the tgger, puts continuous stress on tube an winows but will iniize the heat change probles in the associate accelerator section. - 12 -
Reoving the ve operates the tube in ioe fashion an puts the tube, with respect to stress, between the tgger blocing an outphasing schees. ent. Klystron stanby offers a quic an iscrete for of energy ajust- However, it is not suitable for correction of energy changes ue to continuous paraeter vaations.. Bea-Loaing Effect The relations existing between bea rrent, energy, an analyse rrent are given in M Report No. 2. use by Project M,.1% energy. 4 a change in bea rrent of 1% prouces a change of to a 16 vaation of the analyse rrent. By the efinition of bea loaing With the oel use in the M-2 analysis, this correspons Hence, for heavy bea loaing, substantial control over the analyse rrent can be obtaine with sall changes in bea rrent. bea loaing this for of control is less attractive. For light The schee outline above has the avantage of being siple, con- tinuous an fast. e. Pair-Wise Dephasing When the energy vector in a section is shifte in phase with respect to the bea reference vector, the energy contbution of the accelerator length involve is reuce by an aount corresponing to the cosine of the angle. site irection by the sae aount, the su vector will be in phase with the reference. with. This pair-wise ephasing oes not change the spectru nique can be use to recover the operating energy by suitably reucing the phase angle. The aount of initial ephasing require obviously epens on the ount of energy to be recovere an the nber of sections in- volve # If another vector of the sae length is shifte in the oppo- If an energy loss ocrs, the energy "store away" by this tech- Phase control can be achieve by ferte-type phase shifters with e excitation. the objective. This schee woul be continuous an fast enough to eet A axiu phase shift of up to 6 appears to be feasible with ferte-type phase shifters. When a loss of energy ocrs in the M ' Report No. 2, op.cito
controlle vectors, one vector is shortene an a phase in the su vector is introauee. In this case, the phase vectors assue ifferent angles wlth the reference a the contrci proble is coplicate. situation IS eyeatze when paraeter changes lea to phase changes or to energy chaiges of the inlviual vector in the controlle units, A siilar The following illustration assues the changes to happen outsie the phase ve-,tol-s an necessary ephasiag for an energy range of 1% of the total energy or a loss of one lystron is wore out, A total of 24 tubes are assure for the accelerator with 8 lystrons in each phase vector, of 16 tubes. A 1% change in total energy is a 1% share of the contbution shoul be ephasea by N 3'. In orer tc "store away" that uch energy, each vector by suitably ajusting all or part, of the aceelerator. then available ung operat,:on. one lystron can be ae by a 2 ephasing, 1, BRIEF CaMpARISON AND RAITI'NG OF THE SGHhWS The store-away energy has to be recovere This reserve is Siilarly, provisions for the loss of Obviously, any aount of energy to be use for control or reserve purposes ust coe fro the total energy available. eans are to be provie, there ust be a reserve. Hence, if control With respect, to the viewpoint taen in selecting a suitable energy recovery schee, the ve-level control an stanby schees are rule out. The forer has unesirable sie effects an the latter oes not eet the requireents of being continuous. et in vaous ways an are copare as follows. a. Klystron I Bea-'Joltage I Control The reaining schees have This requires one control point where basic control is alreay pro- vie in the for of a e-q'ing cit control, It is continuous an fast an covers the energy range of 1% an the loss of one tube. It is probable that tube losses up to eight can be recovere if three control points are provie when each control point regulates 32 tubes. b Pair-Wise Dephasing Tn this schee a larger nber of control points is necessary, Con- tinuous an fast control over a Tang of 1% of energy covers energy losses up to o tube, losses ocr in controile units, Control becoes ore coplex when range is extene or - 14 -
e. Bea-Loainn Effects While basically siple, the range is liite an this etho functions well only at high bea rrent, 11. SUMMARY Analysis of the probles ung operation inicates that the analyse bea rrent is expecte to change continuously ue to paraeter vaations such as accelerator teperature, an that the ost frequent source of failure will be the loss of rf power fro one lystron. The latter is ue to failures in lystron an oulator operation. Base on inforation presently available, this is expecte to happen every four hours on the average. Klystron bea-voltage ajustent at a selecte nuber of tubes, e.g., 32, can be use to control the energy vaation cause by the failures entione above. Phase vaation can be offset by a phase control syste acting on the buncher. The require control inforation is, in pnciple, available fro a change in bea energy an bea spectru with. This inforation can be ae available fro seconary electron onitors suitably place across the energy-efining slits. These corrections can be ae autoatically an, when an energy reserve of 1% is provie, lystron losses of up to two can be controlle. This iplies that such losses be r...-i-lz.:e ieiately when they ocr so that, on the average, no ore than 2 lystrons will be out of service at one tie. If such a policy is not feasible, a lystron stanby syste also will have to be provie to allow for greater nubers of lystrons or oulators in nee of repair. Siultaneous losses of larger nubers of lystrons will be rate as single-source failures an will shut own operation while service at the faile unit is restore. The aintenance policy to be use in this case will be a ixture of repair when failures ocr an of preventive aintenance. The selection of a proper policy can be base on ctea such as evelope above in Section 6, which escbes repair of a single-source failure. The repair effort shoul be lai out so that the outage cause is an acceptable fraction of the expeental tie. It is evient that control an aintenance have to be cobine in a suitable way to achieve optiu operating efficiency, i.e., the axiu use of the accelerator for expeental purposes at iniu cost. - 1 -