POTW EXPERT

POTW Expert Expert system to diagnose performance limiting factors at publicly owned treatment works (prototype) ALEX Smalltalk/V Micro L. Berkman/Eastem Research Group ... 

Based on the proven track record and widespread applicability of the CPE/CCP approach, EPA s Office of Research and Development/Center for Environmental Research Information decided in 1988 to fund development of an expert system to incorporate the expertise embodied in the CPE/CCP methodology. This report describes the concept, design, and implementation of this expert system. The system, POTW EXPERT, was beta-tested in September, 1989, and is currently in limited-distribution testing by EPA. 

The CPE/CCP approach to POTW evaluation proceeds in two distinct but linked phases. The Comprehensive Performance Evaluation identifies and prioritizes the design, operational, maintenance, and administrative factors which may be the cause(s) of suboptimal POTW performance. The Composite Correction Program implements procedures (e.g., changes in plant operations, training for plant operators) to address these factors. The CPE and CCP are linked in that implementation of the CCP may feed back to affect the content or order of the prioritized list of factors affecting plant performance. The distinction between CPE and CCP marks a logical problem boundary for an expert system built around the CPE/CCP approach development of POTW EXPERT to date has focused on the Comprehensive Performance Evaluation. 

The objective of the POTW EXPERT development effort is to allow users to rapidly and effectively evaluate the cause(s) of degraded performance at POTWs that fail to meet effluent quality targets. The system achieves this objective by making the proven diagnostic expertise embodied in the CPE/CCP approach available to the community of wastewater treatment regulators and operators responsible for achieving and maintaining POTW effluent quality. 

POTW EXPERT address three types of secondary wastewater treatment processes suspended growth, fixed film, and stabilization ponds. The expert system emphasizes meeting National Pollutant Discharge Elimination Standards (NPDES) for secondary treatment facilities (30 mg/1 B0D5 and TSS) with a maximum average daily flow of 20 million gallons per day (mgd). 

Hardware and Software Requirements and Selection. A number of requirements influenced the early design phases for POTW EXPERT. These were related primarily to the user community for the system, development and delivery hardware, and the need to continually update the system. 

Users POTW EXPERT is targeted to a user community of potentially... 

Platform Because of the nature of the intended user community, POTW EXPERT has been developed for delivery on PC/XT class machines with no more than 640K of memory. (Better performance is obtained, however, if the system is installed on an AT class machine.) Because of the very large knowledge base we anticipated for the system, we attempted to identify an expert system shell with an... 

The development platform is totally open. Specifically, both the ALEX and Small talk/V source code are available to the developer. This feature has allowed us to develop a number of in-house enhancements both to fine-tune ALEX as a development tool for POTW EXPERT, and to improve the performance of ALEX and Smalltalk/V in the PC/XT delivery environment. 

POTW EXPERT S top-level architecture consists of four modules (Figure 1). These are ... 

Figure 2. POTW EXPERT data entry screens tailored to the specific characteristics of the POTW under analysis.

Users need not complete data input in the Data Entry module. Any information omitted by the user will be requested interactively when required by POTW EXPERT during the Major Unit Process Evaluation or Evaluation of Performance Limiting Factors. POTW EXPERT will attempt to reason in the absence of user input data, but will not conclude its reasoning process if critical input is missing. 

Output from this module consists of a ranked listing of the Performance Limiting Factors. POTW EXPERT does not rank PLFs within each Class, but generates a report listing PLFs by Class and providing the information that resulted in each classification. 

Report Generation. Reports generated by POTW EXPERT include ... 

POTW EXPERT S knowledge base is represented in two primary structures ... 

CONTEXTS - - The larger problem of evaluating POTW Performance Limiting Factors is divided into contexts. Contexts define the skeleton of an expert system implemented in ALEX they correspond to relatively isolated sections of the problem under investigation, in our case, to the major evaluation areas of the POTW performance analysis. A hierarchy of contexts can be defined, so that this top level architecture can be subdivided into a number of sublevels. In POTW EXPERT, the top level contexts are system managers, which... 

Figure 3. Smalltalk s hierarchical system of classes and subclasses as implemented in POTW EXPERT.

DEMONS - - Demons are the central knowledge representation structure in an ALEX expert system and are one of the actual repositories of knowledge. In the Smalltalk vernacular, the demon is an object that has a fairly large number of instance variables. A demon corresponds to each of the areas in which a cause of a POTW performance problem may be isolated in POTW EXPERT, at least one demon has been defined for each of the 66 potential Performance Limiting Factors identified by Process Applications. 

Each demon accumulates evidence that its associated PLF is or is not a factor that limits plant performance. This evidence is represented by the demon s level of current belief, which is determined by the effect of the individual variables related to the demon. Many demons may "observe" the same variable, but each demon will have a distinct reaction to the value of the variable. For example, resolution of a variable which increases the level of belief in one demon can simultaneously decrease the level of belief in one or any number of other demons. A critical design component of an ALEX expert system, therefore, is the definition of which variables will be observed by each demon, and determination of the impact each variable will have on the belief of each associated demon. Once a demon acquires control of a system consultation, investigation of variables related to the demon continues until all variables related to that demon have been evaluated. At the end of the consultation, the final level of belief of the demons also defines the system s conclusions -- with POTW EXPERT, the PLF demons final levels of current belief determine whether or not each PLF is judged to be a cause of degraded plant performance. 

The POTW EXPERT system demonstrates that a microcomputer-based expert system can effectively represent a complex evaluation methodology, evaluate the capability of a secondary wastewater treatment facility s major unit processes, detect factors which potentially limit performance, and categorize them according to their influence on plant performance. The model is presented in a logical and structured manner to allow wastewater professionals unfamiliar with the CPE process to effectively employ this wastewater treatment methodology. 

Development of POTW EXPERT has been funded by the United States Environmental Protection Agency under Contract 68-C8-0014, Work Assignment No. 0-01. The contents of this article, however, reflect the views of the authors and do not necessarily reflect those of the U.S. Environmental Protection Agency. Mention of trade names, products, or services is not, and should not be interpreted as conveying official EPA approval, endorsement, or recommendation. 

---