Page 95 - Improving Machinery Reliability
P. 95

Vendor Selection and Bid Coriditionirig   67

                  Mechanical Seal Selection and Evaluation

                    Pump-application engineers will generally agree that seal and seal environmental
                  system selection  on  many pumps  is becoming more complex  and time consuming
                  than pump selection.’2 Still, an average of only 10%-35% of the allotted pump-engi-
                  neering time is generally spent on the mechanical seal system.13 Looking at the cost
                  of seal failures and such consequences as major fires, release of  toxic materials, and
                  unit downtimes, there is reason to believe that more time should be spent on seal sys-
                  tems design. But before we decide who should spend this time, we should examine
                  the various selection practices prevalent in the petrochemical industry.
                    Seal  selections  made  entirely  by  the pump  vendor  have  generally  proven  to be
                  least reliable. The pump vendor is concerned that his competitor will underbid him,
                  and thinks that the engineer selecting the pump will only look at the initial, installed
                  cost without giving credit to the potential run-length extension and maintenance cost
                  avoidance of  superior  seal components or seal system designs. Consequently, the
                  least expensive seal is often selected, leaving plant operations or maintenance bur-
                  dened with  an inherently weak  seal. Furthermore, pump manufacturers  seldom
                  receive experience feedback on seals furnished with their pumps. Seal selection by
                  the pump vendor alone should thus be discouraged.
                    Contractor’s or user’s standards have generally been applied with somewhat high-
                  er success. Unfortunately,  many of  these standards are full of  generalities and give
                  little guidance on specific requirements.  Very often the stated requirements do not
                  separate barely acceptable from truly successful seal systems. Lack of specific guid-
                  ance in an otherwise well-intended specification may significantly impair its useful-
                  ness and deprive the user of a low-risk sealing system.
                    Optimum seal  selection  practices  should  make  extensive use  of  vendor  experi-
                  ence. These practices must encourage  the  seal vendor to use his own gland design
                  and to recommend seal systems, not just seals. To properly advise the user, seal ven-
                  dors require full information on product composition, process conditions, crystalliza-
                  tion temperatures, solids entrainment, and the like. All of these data are highly rele-
                  vant if proper selection is to be ensured, and withholding data for “security” reasons
                  may cost the user dearly. Optimum seal selection consists of the following steps:


                    All relevant data must be disclosed to the seal vendor. If  security is truly a valid
                    concern, disclosure should be preceded by signing confidentiality agreements.
                    At  least three and preferably four major seal manufacturers with strong and capa-
                    ble representation  in the  user’s  geographic  location  should be  invited  to  submit
                    bids. The user should screen and verify vendor capability by such criteria as ability
                    to furnish  engineered seal components, e.g.,  special  pumping  screws  instead  of
                    ineffective pumping rings, and by  vendor’s willingness to stock appropriate spares
                    in the user’s geographic area.
                    The bid  invitation should clearly state that the user is interested in buying  a seal
                    sysrem, not just a seal.
   90   91   92   93   94   95   96   97   98   99   100