User Group Newsletter - nHance Tech

B&W/MMSUser GroupNewsletter

MMS: A COMPUTER PROGRAM DEVELOPED BYTHE ELECTRIC POWER RESEARCH INSTITUTE

September 1985 Vol. 1 No. 2

The President's Corner

I have been pleased with thedevelopment of the B&W/MMS UserGroup over these first six months. Ithas been successful from both thesupport given by B&W to the usersand the contributions made by theusers in the form of suggestions forimprovements.

Based on the concerns that weusers expressed at the first UserGroup meeting, B&W will proposesome modifications to the member-ship agreement. Phil Bartellswill bereporting on these at our Septembermeeting.

Another area where users havemade a worthwhile suggestion is aUser Group electronic bulletinboard. Hemmat Safwat of Bechtelwrote a letter to B&W with somesuggestions for several improve-ments including the use of the elec-tronic bulletin board. B&W has nowset up such a bulletin board andPhil will show us how to use it at theSeptember meeting.

In July I represented the B&W/MMS User Group at a meeting be-tween EPRI and B&W in Palo Alto.

The purpose of this meeting was toreview where B&W stood on the listof tasks they were required to do aspart of their license agreement withEPRI. It appeared that B&W hadaccomplished all of the major tasksfor EPRI with only a little moredocumentation to be done. It shouldbe remembered that the EPRI list oftasks differs from the prioritized listwe users put together at the pastUser Group meeting. We will beholding B&W responsible to our list.

The membership in the B&W/MMS User Group is growing. I wouldlike to formally welcome the follow-ing organizations which have joinedsince the March User Group meet-ing:• Cleveland Electric Illuminating• Utah Power and Light• Arizona Public Service• Empresarios Agrupados, S.A.

As I see it, the more members wehave in the User Group, the moreinterchange of MMS experience wecan have and the more improve-ments to MMS can be made byB&W.

Charles SaylesPresident

August 29, 1985

EPRI Completes Development of Air/Gas DynamicsThe EPRI project to develop en-

hanced variations of the MMS fossilmodules to simulate the pressure/flow dynamics on the air/gas sideof the boiler modules is due to becompleted in the early Fall of thisyear. These modules are intendedto provide users the capability tosimulate startup, shutdown, andoperation of the draft and firingsystems in fossil fired units. Thiscapability is not available in thepresent version of MMS.

Module formation is firmly basedon rigorous application of mass andenergy conservation. Combustionanalysis is based on the results of

the EPRI DUCSYS code and 150years of B&W boiler design experi-ence. These modules will add thecapability to analyze the draft andcombustion control systems to theexisting capability to analyze thefeedwater system, steam system,and control loops to provide thecapability for analysis of the entireplant system.

The new modules may be invokedinstead of current modules AIRHX,DRUMFC, DRUMNC, ECON, OTBLR,PULV, REGHX, and SPRHTR. Thenew versions of these modules,AIRHXA, DRUMFA, DRUMNA, ECONA,OTBLRA, PULVA, REGHXA, and

SPRHTA, will be supplemented bymodules representing the combus-tion components (BURNER), connec-tive modules (CONNA), dampers(DAMPER), dividers (DIVA), ducts(DUCTR, DUCTRS, and DUCTSR),axial fans (FANASR), centrifugalfans (FANCSR), junctions (JUNCA).These new modules, and the modifi-cations of the existing modules,follow the same Resistive-Storagemethodology upon which MMS isbased. Upward compatability is as-sured by retention of the current,quasi-steady-state fossil moduleswithout change.

- LANCE SMITH

Steady-State Finder Evaluation ContinuesEvaluation of alternatives to the

current ACSL TRIM steady-statefinder routine continue. Reports ofproblems with the TRIM routine area familiar litany to experienced MMSusers. Difficulties are particularlyencountered as the size of the modelapproaches 50 states or more. Re-sponse to these problems has re-suited in several analyses by EPRI,Mitchell & Gauthier, B&W, andArgonne National Laboratory (ANL)to identify the cause and provide afix. Recent studies by B&W haveanalyzed the current TRIM routineand a developmental Mitchell &Gauthier routine. In a separatestudy, Dr. David Hetrick at the Univer-sity of Arizona is applying anothertechnique to steady-state finding forMMS models.

The current ACSL TRIM routineuses a Newton-Raphson techniqueto iteratively update the states in anattempt to reduce the square rootof the sum of the squares of thederivatives below a threshold value.In order to apply the TRIM routine,which is based on solution of linearproblems, to the highly non-linearMMS problems, the user is requiredto manually adjust a parameterwhich controls the fraction of thelinear step taken at each iteration.

Under EPRI funding, Dr. HowardGeyer of ANL developed a hybridtechnique which utilizes a combina-tion of a Newton-Raphson and aSteepest Descent method. The Steep-est Descent method calculates thedirection and distance to update thestates to minimize the square rootof the sum of the squares of thederivatives. It is this calculation ofthe distance, which relieves the userof any manual adjustment of param-eters, which present the major ad-vantage of the technique. B&W testsof the hybrid technique suggest thatit is no poorer than the current ACSLTRIM routine for large models andrequires fewer User inputs for small-er models.

Mitchell & Gauthier are committedto incorporate a modified version ofthis steady-state finder in the nextrelease of ACSL. B&W has alsotested this version of the algorithmand made recommendations for itsuse. With these, it should be pos-sible for MMS Users to utilize thehybrid technique or to fall back onan algorithm which is essentiallyidentical to the current routine.

The B&W analysis indicates thatmany of the previous problemswhich have been assigned to thesteady-state finder are, in fact, more

accurately attributable to either themodel or to peculiar characteristicsof some MMS modules. B&W hasbeen able to TRIM the 100+ stateTMI turbine trip model to a steady-state with the current routine byessentially eliminating the pressur-izer module states. An analysis sug-gests that the high non-linearitiesassociated with the equilibrium/non-equilibrium transition presentalmost insurmountable obstacles tothe steady-state finder. This analysisalso suggests that even small modeldefinition problems can completelyincapacitate the steady-state finder.In the course of the study, B&Winadvertently applied the new ACSLtechnique to a 26-th order modelwhich included a string of com-ponents with flowrate boundary con-ditions at both the upstream anddownstream ends with non-identicalflowrates as boundary conditions.The steady-state finder self destruct-ed with few indications of the fun-damental problem. A short integra-tion of the model resulted in rapidlydecreasing pressures in a specificportion of the model which resultedin the identification of the problem.Together with previous experiencewith models with small parameteriza-tion errors which prevented the

model from conserving mass orenergy, this suggests that no steady-state finder is a good tool to use tobring an un-debugged model tosteady-state. B&W's conclusion isthat this is the cause of many of thepreviously reported problems withthe TRIM function. This result isconsistent with the limited successthat some of the user communityhave had with the current routine.

Additionally, application of a stilldifferent routine is currently being

analyzed at the University of Ari-zona. Under contract to B&W, Dr.David Hetrick is applying a patternsearch routine to two MMS modelssupplied by B&W. Work to date isinconclusive, but has demonstratedthat the technique, called INCON,can be applied to MMS models.

The search continues, but resultsnow suggest that many of the prob-lems reported previously may nothave been completely attributableto the ACSL TRIM routine, but to

inherent model problems resultingfrom either parameterization errors,inconsistent boundary conditions, ormodule formulation. It is clear thatno steady-state finder can reason-ably be expected to find a steady-state for a model which has nosteady state. The best that can beasked of this routine is that it diegracefully and in a manner thatsuggests where the fundamentalproblem lies.

- LANCE SMITH

Extended Range Features ImplementedThe Extended Range Features con-

sist of a set of modules which havebeen modified to provide additionalcapabilities. These modificationsallow the modules to calculate sys-tem dynamics during conditions of:

• Low flow• Reverse flow• Two-phase flow

In implementing these capabilities,the model ing a p p r o a c h hasbeen to provide a calculation mecha-nism that works simply and reliablyrather than to provide a high degreeof detail in the modeling. The pri-mary objective was to permit calcu-lation through a short, relativelyunimportant excursion by a modelinto these regions.

The following lists the moduleswhich have been modified to includethe Extended Range Capabilities:PIPER CONN1 AIRHXPIPERS CONNC ECONPIPERSR RX1 DRUMFCPUMP UTSGA DRUMNCVALVEC UTSGE OTBLRVALVED OTSGEM FWHTRVALVE1 PUMP4Q FLASHJUNC PZRB CONDENDIV SURJNC SPRHTR

The status of this development isthat all modules have been modifiedand tested. Testing has includedboth single module tests and systemtests. For the system tests, a fossilboiler model and a B&W nuclear

system primary model were con-verted to use the modified modules.These models were tested to seethat the modules produced the sameresults as the conventional modulesin normal operation as well as test-ing the new features. The extendedrange modules are expected to bereleased as additional modules inthe MMS library. Names will be mod-ified for the new modules so thatusers may use both old and newversions of the modules. Interfacemodules are also being developedto connect conventional and extend-ed range modules in the same model.

- TOM WILSON

MMS-EASE+ Version 0.91 Released to UsersVersion 0.91 of the MMS-EASE+

pre- and post-processor packagewas delivered to eight member or-ganizations during July and August.This version has pre-processing forall the BOP modules, all the Fossilmodules, and the following nuclearmodules: MSRHT, PZRB, RX1,SURJNC, UTSGA and UTSGE. Pre-processing activities include modelbuilding using interactive graphics,on-screen forms for data entry anddisplay, automated parameterizationcalculations, automated water prop-erty look-ups and automatic prepara-tion of model files and commandfiles. The model files may containuser-defined macros and/or FORTRANsubroutines, along with user-speci-fied inline ACSL coding. The com-mand file contains a user-customized

PREPAR list that is generated as themodules are parameterized. Withoutexiting from MMS-EASE+, the usermay call up any communicationsprogram for sending the model andcommand files to a mainframe.Post-processing activities involve ex-tracting plot data from ACSL outputfiles and generating plots using inter-active graphics.

Additional features and moduleswill be added to the MMS-EASE+system in future versions. Some ofthe additional features currentlyplanned are:• Capability for larger models by

linking together multiple model"pages."

• Display of results on the modeldiagram. This function will dis-play initial conditions and per-haps eventually be capable of

displaying transient results (pres-sures, temperatures, etc.) on theappropriate flow stream.

• Linking with MMS/PC. See thearticle on MMS/PC for moredetails.

• A MMS-EASE+ report generator,which will provide the documen-tation for MMS-EASE+ developedmodels with minimal user ac-tion.

• A more user-friendly plottingpackage.

• Extended capability for pre-pro-cessing of controls-type mod-ules.

• Optional high resolution colorgraphics capability.

- ROSS SCHAACK

MMS/PC - A New Way to Run MMSWith the advent of the MMS-

EASE+ pre- and post-processor andACSL/PC, it is time to bring MMS tothe PC. MMS/PC uses the sameACSL macros as the mainframeversion, and (functionally) the samewater properties routines. It requiresthe ACSL/PC translator, availablefrom Mitchell and Gauthier Asso-ciates (MGA), and the Microsoftversion 3.2 FORTRAN compiler. Thecurrent version of MMS/PC is limitedto very small models (less than 20states, unless only a few modulesare used, then as many as 30 states),and it is not as accurate in itscalculations as mainframe MMS.However, for small models, MMS/PCis probably a more efficient tool thanmainframe MMS.

MMS/PC has a different sort oflibrary from the mainframe MMS

library. The macro coding is usedas the MMS library, and a programsupplied with the MMS/PC packageis run to insert the required MMSmodules into a copy of the modelfile. The library was handled in thisway so that the limited amount oftable space currently available inACSL/PC could all be used for theuser's model, rather than for modulesthat are not called.

MMS/PC can be linked to theMMS-EASE+ pre-processor so thatthe user does not have to exit fromMMS-EASE+ to run his model. Thetranslation and compilation processis fairly time-consuming (20 minutesfor the deaerator level control studymodel), but the entire process canbe done without user attention, sothe user can be accomplishing othertasks while the PC is doing its pro-cessing.

ACSL/PC has all the commandsand capabilities of mainframe ACSL,so the MMS user will be right athome with MMS/PC. Command filescan be swapped between machineswithout modifications. Also, ACSL/PCcan produce plots, giving theMMS/PC user another option forpost-processing activities.

B&W has recently obtained thesource code for ACSL/PC under aconfidential arrangement with MGA.The next step is to attempt to modifyACSL/PC so that it can utilize allavailable memory in the PC for tablespace, thereby dramatically increas-ing the model size that can betranslated. If all goes well, it isplanned to offer this improved ver-sion of MMS/PC by the end of 1985.

- ROSS SCHAACK

Gasification Combined Cycle Analysis May Improve MMSEPRI has awarded Stone & Web-

ster Engineering Corp. (SWEC) andB&W a contract to develop a model-ing system for gasification/combin-ed cycle processes.

In general, the GCC modeling ap-proach will follow the MMS formatsand technical approach. However,the wide range G'CC applicationsimpose some special requirementsthat the MMS approach does noteasily accommodate:1) The GCC models will be run on

small computers in realtime forsome applications as well as thetraditional mainframe computerapplications. MMS is not cur-rently designed to run on smallcomputers in realtime becauseof the numerical integration meth-ods used.

2) GCC models are likely to belarger than typical MMS modelsthereby requiring substantialmemory and table space forcompilation. The MMS approachcurrently uses a large amountof table space. Therefore, theGCC approach must avoid un-necessary use of table space toallow substantial size models.

3) MMS was designed primarily forsteam/water systems. The GCCmodels will contain over 15 chem-ical species which requires amodification to the MMS nomen-clature.

To alleviate these limitations, thefollowing modifications to the MMSapproach are planned for the GCCmodeling system:1) The use of connective nodes for

linking two resistive type mod-ules in series will be eliminated.This will reduce the number ofvariable names in a given modeland alleviate the numerical in-tegration problem caused byconnective nodes. This will bereplaced by a "netnode" thatwill calculate an equivalent resis-tance of the resistive modulesin series automatically.

2) Pressure calculations at flowsplits and merges will be formu-lated to allow an integration stepsize of 1 second rather than thevery small step size currentlyimposed by connective nodes.

3) The GCC module coding will usesubroutines more extensivelythan currently used in MMS.

This will substantially reducecomputer memory require-ments, and translation time re-quired.

4) The MMS nomenclature will bemodified to accommodate anynumber of chemical constitu-ents. This will be done by usingarray variables instead of scalarvariables. This approach alsohelps to reduce the table spacerequirements of the final model.

If successful, the above changescould be expected to be incorporat-ed into the MMS-B&W at some laterdate. This would require approvalby EPRI in order to take place.

- LANCE SMITH

MMS User Group NewsletterEditors: Phil Bartells (804) 385-2637

Paul Henningson (804) 385-2209

Published quarterly by Babcock & Wilcoxfor the MMS User Group to report on currentMMS developments and User Group ac-tivities.