P.Cornwell and J.Thrower
This report describes the development of a sand vacuum clamp tailored to the needs of the small foundry and studio. Ceramic shell moulds are put under considerable stress during casting and a reliable means of clamping the mould and supporting its external surface is imperative. The sand clamp achieves this without impeding the flow of gases through the porous shell wall during metal pouring. It reduces the risk of cracking and significantly improves safety.
Although sand vacuum clamps are available commercially, they are expensive and require commercial infra-structure beyond the scope of most studio environments. The aim of the work reported here was to develop a low-cost alternative and to better understand the clamping of ceramic shell moulds, in line with the goals of the Central Saint Martins Foundry Research Project, to make such techniques more widely assessible.
Ceramic shell moulds offer many advantages including preserving fine surface detail, porosity leading to greatly simplified mould design by reducing the need for risers, and ease of construction and handling. However, special attention is necessary to limit stresses on the fragile shell during casting. The sand vacuum clamp is ideal for this purpose because of its ability to provide uniform support of the shell surface and replace the need for most external bracing, which otherwise causes problems of differential thermal expansion.
The reliable and reproducible casting process which is emerging from this work is based on good information about and control of conditions within the sand body. This note describes the basic operation of the clamp, it summarises work undertaken by David Reid, with the assistance of Anderson Inge, to investigate clamping pressure and it discusses further research in progress at the end of July 1996 and plans for development of a new clamp design based on our results.
Ceramic shell moulds are used to make metal casts by the lost wax method. The wax original may be unique and will always involve a considerable amount of work. The shell is built up in several layers which must each dry before de-waxing and subsequent vitrification can take place to produce a finished mould. Failure during casting is therefore generally very costly as well as being potentially dangerous. When molten metal is poured into a shell stresses arise from the rapidly changing distribution of mass of metal, from thermal expansion and from internal pressure due to expanding gasses. These problems increase rapidly with larger scale work.
The sand box vacuum clamp provides a solution by supplying uniform mechanical support over the whole surface of the shell and consequently, the shell can be made thinner. This saves construction time and results in a mould which is more porous and, for that reason, more easily purged of trapped air and gasses from the molten metal. This significantly reduces the need for risers and in turn, further reduces construction time and fettling of the cast. The sand clamp ensures that the mould will not topple over during pouring and it contains explosions and metal spills if the mould does fail. Further, the flow of air through the sand body allows some control of the rate of cooling of the mould which effects distortion, metal surface quality and strength.
Traditional mechanical backup methods - for strengthening ceramic moulds with glass fibre or refractory cements - result in the introduction of stresses due to uneven expansion during the rapid temperature changes accompanying metal pouring. The sand clamp eliminates these problems and reduces remaining expansion stresses by allowing thinner shells to be used.
The sand vacuum clamp consists of an open steel vessel containing silica sand. At the bottom, divided from the sand body by a filter membrane is a plenum chamber. The plenum can be evacuated or pressurised by means of external pumps, such that air pressure in the sand body can reduced below that of the surrounding atmosphere or air can be blown up through the sand. By creating a partial vacuum in the plenum, air is drawn through the sand to aid removal of gases driven through the wall of the ceramic shell. When 'fluidised' with compressed air, it is possible to insert fragile shell moulds into the sand with negligible force.
The filter membrane provides mechanical support of the sand, prevents all but very fine particulate being carried in the airflow to the vacuum pump and controls distribution of fluidising air through the sand.
Clamping pressure is achieved by agitating the vessel with a vibrating device to compact the sand body.
The containment vessel of the current Foundry Project clamp consists of an open one metre cube box. Its sides are made of 3mm sheet steel, reinforced at the corners, around the plenum filter and the vibrator mounting with 5x40 mm angle section steel. The vessel is 80% filled with equal parts of 65 and 110 grade silica sand and the structure is supported on four car tyres on the foundry floor to gain maximum compaction of the sand from an electrical vibrator. The latter is a 3- phase Vibramotor unit of 550W, made by Viking, producing 6KN, which is bolted directly to one side of the vessel.
The plenum filter support is fabricated from steel mesh supported on 40mm angle section welded to the wall of the vessel. It is covered with a filter membrane made of 500g/sq metre canvas, sealed with silicone rubber compound. Two pipes are welded to the plenum wall below the filter and are connected through flexible couplings to the pumps. The vacuum pump used in the production sand clamp at Central Saint Martins is an NGN rotary oil type RP069 three phase pump of 2.2KW which, with the current geometry of the plenum, filter and sand body, maintains a vacuum of 30mBar, measured at the plenum airway. A secondary oil-impregnated in-line filter in this airway protects the vacuum pump by trapping particulate drawn from the sand-body through the primary canvas filter. A valve isolates the secondary filter and pump from the plenum to prevent ingress of sand during fluidisation. Inspection has revealed negligible granular sand having been drawn through the filter membrane during twelve months of regular use, without requirement to change the secondary filter during that period.
A manual valve on an airline maintained at 6Bar supplies air for fluidisation of the sand. The valve is adjusted to prevent large bubbles building up in the sand body and propelling sand from the clamp. This can be dangerous when the sand is hot from previous casting since it can cause burns and at all times it represents a respiratory hazard. It is not possible to prevent fine silica from being carried into the working environment by the airstream during fluidisation and dust and fume precautions are necessary.
1. Fluidisation. Compressed air is introduced into the plenum and exhausted through the sand body for just long enough to insert moulds. This unlocks the sand particles and allows ceramic shells to be introduced by hand without damage. Distribution of fluidisation air throughout the sand body can be controlled by varying the porosity of the canvas filter across its surface with paint markings and becomes more critical with larger moulds. Further research is necessary in this area to determine suitable paint types and filter markings in relation to the entry point of the airway into the plenum and the cross section of the vessel, in order to prevent 'hard spots' of low fluidisation that may damage large moulds during insertion.
The pouring cup of each mould should project a few centimeters above the sand surface. Care must be taken to prevent sand entering the mould, which results in voids in the cast. This is particularly likely during fluidisation and a kaowool plug and aluminimum foil cover should protect the pouring cup until immediately before casting.
With repeated use, the sand becomes hot enough for fluidisation to become an important factor in control of cooling in the critical period of solidification (airflow resulting from plenum vacuum is also important in this context: see below).
2. Compaction. The vibrator is typically energised for a period of 10s, which compacts the sand body by causing the sand granuals to re-orient and lock together. This action ensures uniform clamping pressure on the shell surface and resists sudden expansion during metal pours, especially eruption around fractures. Slower thermal expansion of the shell is accommodated. It is important that the sand body does not contain solid impurities, or in the event that multiple shells are clamped, that they do not collide during compaction since shell failure invariably results.
While not directly proportional to the compaction time, clamping pressure increases most significantly during the first 10s of operation of the vibrator. The settling of the surface level of the sand body from that following fluidisation provides a rough measure of compaction and clamping pressure, but this alters during application of vacuum. Low compaction, even with vacuum, results in failure of the mould during metal pouring, while extreme compaction can cause failure during clamping, effectively crushing shell walls. Between these limits, adjustment of the compaction time can prevent ingress of particulate through small fissures in the shell, leading to voids in the cast, and avoid shell failure during metal pouring.
Considerable work has been undertaken in the Foundry Project to determine factors effecting clamping pressure and its relationship with shell failure. Quantitative work has been undertaken by Anderson Inge and David Reid using a test rig in which conditions can be controlled more easily than in the main Foundry clamp and preliminary results are presented in Table 1. Further research involving electronic instrumentation of the sand body to measure actual clamping pressure and also investigation of alternate vibration techniques is in progress.
3. Vacuum Clamping. Vacuum is applied to the plenum after compaction and is sustained during metal pouring. Vacuum effects clamping pressure by altering the locking between sand granuals and its precise contribution to clamping is the subject of further work. Plenum vacuum can be increased for a given vacuum pump, by covering the top surface of the sand with a non-porous membrane such as polythene sheet. Atmospheric pressure is then more effectively transferred to the surface of the sand body, although the effect upon clamping pressure of mechanical loading of the sand surface with a baffle plate and weights is also being investigated.
A related but equally important effect of vacuum applied to the sand body is gas movement induced through the porous shell wall. This reduces the tendency for blow back in the sprue during pouring but also evacuates air trapped by inverted concave spaces. Consequently risers are seldom required for shells used in the vacuum clamp and even where employed, they are usually blind. Further, fumes produced by residue remaining within the shell at pouring time, and even by ablative direct casting material such as polystyrene, are evacuated through the shell wall and purged by the action of airflow drawn from the atmosphere to the pump through the filter. This side effect is particularly valuable where toxic fumes are produced, since the exhaust from the pump can be further processed and venting controlled, in contrast to fumes escaping into the working environment.
Finally, the vacuum airflow also has a marked cooling effect after pouring, in the critical period of solidification. Control of the cooling profile is the subject of important future research, since both contraction of the casting, associated accuracy of surface detail, crystalline structure, surface finish, mechanical strength and surface hardness are dependent upon this.
4. Release and Preparation. Solidified casts can be withdrawn by hand from the sand. Fluidisation for a period after release modifies the temperature of the sand body, rapidly distributing heat from the lower volume around the thermal mass of the released cast towards the surface and cooling the sand overall. In combination with dummy casting to pre-heat the sand body, considerable control can be exercised over shell temperature at pouring time and as above, during cooling around solidification. Further research in this area will follow current work analysing data on clamping pressure.
To overcome difficulties of non-uniform compaction and clamping pressure due to debris in the sand body and the geometry of the production sand clamp, a test rig has been constructed to allow investigation of shell failures under controlled conditions. This apparatus consists of a 55 gallon drum, open at one end and 80% filled with the same composition of silica sand as the production clamp described above. A removable plenum assembly is used, lowered into the drum before filling with sand and connected to compressor and vacuum pumps via airways rising through the sand. Sand compaction is achieved by rocking the drum on a rod fulcrum to impact the concrete floor so that compaction due to a given number of impact cycles, following thorough fluidisation, is very reproducible.
A test instrument, designed by Reid and Inge , has been developed for investigation of clamping pressure and shell failure. This comprises a rigid enclosure in two halves, containing a sealed bladder that can be inflated gradually to known air pressures. Electrical contacts located on the mating edges of the halves of the enclosure trigger an indicator lamp when the air pressure within the bladder forces the enclosure apart against the clamping action of the sand. When immersed at a given depth in the sand body, in the same way as a ceramic shell, the instrument provides an air pressure reading directly proportional to clamping pressure at that location. Consequently, relative clamping pressure can be determined against the degree of compaction of the sand in terms of number of impacts.
Table 1. shows the results of tests conducted by this method and demonstrates the consistency of clamping pressure achieved with compaction. This corresponds to average settlement of the surface of the sand body of 10mm for 5 impacts, 17mm for 20 impacts, 35mm for 40 and 40mm for 60 impacts. Further compaction, however, results in marginal increase of clamping pressure and while many ambitious and technically difficult casts have been achieved using the clamp design described, further research is in progress to determine the effect of mechanical loading of the surface of the sand body. In particular, the distribution of external loads to the sand body around a shell is being investigated using electronic pressure measuring apparatus.
The test rig itself is also the subject of further development. Compaction using an air powered vibrator is being evaluated to allow a very low cost and controllable sand clamp to be produced.
Table 1.
Compaction Impacts
0
5
10
20
40
60
Instrument Separation Pressure
~2.0
~4.5
~5.2
~6.8
~7.8
~9.5